Merge mai/hades/styles-expansion: add 6 style presets (cross-coord with flexsiebels)

Cross-coordination with flexsiebels/head (paul). m wants more style options
on /imagine/new; flexsiebels has the UI side ready to bump IMAGEN_STYLES
in lib/server/imagen.ts + schemas.ts as soon as the worker accepts them.

styles.yaml: 6 new entries with FLUX-friendly prompt fragments. No code
changes — Apply() and Styles() consume the embedded YAML directly, the
"enum" is dynamic.

prompt_test.go: extend TestStylesContainsAllExpected expectation list
(alphabetical, '3' < 'a' so 3d-render leads).

Total enum: 11 (5 existing + 6 new). flexsiebels delegation message 1669.

CLAUDE.md

@@ -10,13 +10,17 @@ and lifecycle of its own block in `~/.config/imagen.yaml`.
 ## Architecture
 
 ```
-cmd/imagen/                CLI shell — generate, backends, config, serve
+cmd/imagen/                CLI shell — generate, worker, backends, config, serve
 internal/backend/          Backend interface + Registry + Mock reference impl
 internal/prompt/           Style preset registry (embedded styles.yaml)
 internal/output/           Filename templating, image writer, JSON sidecar
 internal/config/           YAML loader, validation, sample generator
+internal/cloud/            Supabase Storage + imagen.images writer
+internal/usage/            mai.imagen_usage cost-tracking sink
+internal/worker/           imagen.jobs queue consumer (DB-agnostic via Queue interface)
 internal/server/           HTTP stub (not implemented yet — follow-up issue)
-docs/                      architecture.md, usage.md
+scripts/                   imagen-worker.service + env template, ComfyUI scripts
+docs/                      architecture.md, usage.md, setup-worker-mriver.md
 ```
 
 Data flow for `imagen generate`:

cmd/imagen/compare.go

@@ -0,0 +1,386 @@
+package main
+
+import (
+	"context"
+	"encoding/json"
+	"flag"
+	"fmt"
+	"image"
+	"image/color"
+	"image/draw"
+	"image/png"
+	"io"
+	"os"
+	"path/filepath"
+	"sort"
+	"strings"
+	"time"
+
+	"golang.org/x/image/font"
+	"golang.org/x/image/font/basicfont"
+	"golang.org/x/image/math/fixed"
+
+	"mgit.msbls.de/m/ImaGen/internal/backend"
+	"mgit.msbls.de/m/ImaGen/internal/config"
+	"mgit.msbls.de/m/ImaGen/internal/output"
+	"mgit.msbls.de/m/ImaGen/internal/prompt"
+)
+
+// runCompare implements `imagen compare "<prompt>" --models a,b,c --output <dir>`.
+//
+// Each backend in --models runs sequentially against the same prompt (mRock
+// has a single GPU; parallelising would just OOM). Each generation lands as
+// a backend-suffixed file in the output dir; a contact sheet stitches them
+// together into one PNG with the backend name overlaid on each cell. A
+// sidecar JSON next to the contact sheet lists every generation with its
+// per-model metadata (latency, seed, model file, VRAM peak).
+func runCompare(ctx context.Context, args []string) error {
+	fs := flag.NewFlagSet("compare", flag.ContinueOnError)
+	var (
+		modelsCSV  string
+		size       string
+		outDir     string
+		style      string
+		negative   string
+		seed       int64
+		steps      int
+		configPath string
+		noContact  bool
+	)
+	fs.StringVar(&modelsCSV, "models", "", "comma-separated backend instance names (required)")
+	fs.StringVar(&size, "size", "1024x1024", "WxH for every backend")
+	fs.StringVar(&outDir, "output", "", "directory to write the images + contact sheet (default: ~/Pictures/imagen/compare)")
+	fs.StringVar(&style, "style", "", "style preset applied to the prompt before dispatching to each backend")
+	fs.StringVar(&negative, "negative", "", "negative prompt (forwarded to every backend that supports it)")
+	fs.Int64Var(&seed, "seed", 0, "deterministic seed for every backend (0 = each backend rolls its own)")
+	fs.IntVar(&steps, "steps", 0, "diffusion steps (0 = each backend's default)")
+	fs.StringVar(&configPath, "config", "", "config file path (default: ~/.config/imagen.yaml)")
+	fs.BoolVar(&noContact, "no-contact-sheet", false, "skip the composite PNG; only write per-backend images + sidecar")
+	fs.Usage = func() {
+		fmt.Fprintln(fs.Output(), `Usage: imagen compare "<prompt>" --models a,b,c [flags]`)
+		fs.PrintDefaults()
+	}
+	leadingPositional, flagArgs := splitLeadingPositional(args)
+	if err := fs.Parse(flagArgs); err != nil {
+		return err
+	}
+	positional := append(leadingPositional, fs.Args()...)
+	if len(positional) == 0 {
+		fs.Usage()
+		return userErr("missing prompt")
+	}
+	rawPrompt := strings.Join(positional, " ")
+	modelNames := splitCSV(modelsCSV)
+	if len(modelNames) == 0 {
+		return userErr("--models is required (comma-separated backend instance names)")
+	}
+
+	w, h, err := parseSize(size)
+	if err != nil {
+		return userErr("bad --size: %v", err)
+	}
+
+	cfg, cfgErr := config.Load(configPath)
+	if cfgErr != nil && !os.IsNotExist(cfgErr) {
+		return cfgErr
+	}
+
+	if outDir == "" {
+		home, _ := os.UserHomeDir()
+		outDir = filepath.Join(home, "Pictures", "imagen", "compare")
+	}
+	outDir = config.ExpandPath(outDir)
+
+	finalPrompt, err := prompt.Apply(rawPrompt, style)
+	if err != nil {
+		return userErr("%v", err)
+	}
+
+	runID := time.Now().Format("20060102-150405")
+	runDir := filepath.Join(outDir, runID+"-"+output.Slug(rawPrompt))
+	if err := os.MkdirAll(runDir, 0o755); err != nil {
+		return fmt.Errorf("mkdir %s: %w", runDir, err)
+	}
+
+	results := make([]compareResult, 0, len(modelNames))
+	for i, name := range modelNames {
+		fmt.Fprintf(os.Stderr, "[%d/%d] %s ...\n", i+1, len(modelNames), name)
+		res, err := generateOne(ctx, cfg, name, finalPrompt, negative, w, h, seed, steps, runDir, rawPrompt)
+		if err != nil {
+			// Don't abort the whole run on a single backend failure — record
+			// the error and continue. flexsiebels-style consumers want to
+			// see N-1 results rather than zero when one model is offline.
+			fmt.Fprintf(os.Stderr, "    failed: %v\n", err)
+			results = append(results, compareResult{Backend: name, Error: err.Error()})
+			continue
+		}
+		fmt.Fprintf(os.Stderr, "    %s (%d ms)\n", res.ImagePath, res.LatencyMs)
+		results = append(results, res)
+	}
+
+	// Sidecar JSON beside the run dir captures every attempt.
+	sidecar := filepath.Join(runDir, "compare.json")
+	if err := writeCompareSidecar(sidecar, rawPrompt, style, negative, w, h, seed, steps, results); err != nil {
+		return err
+	}
+	fmt.Fprintln(os.Stderr, "sidecar:", sidecar)
+
+	// Contact sheet stitches the successful results together. If every
+	// backend failed there's nothing to draw, so skip silently.
+	if !noContact {
+		successes := successfulResults(results)
+		if len(successes) > 0 {
+			sheet := filepath.Join(runDir, "contact-sheet.png")
+			if err := writeContactSheet(sheet, rawPrompt, successes); err != nil {
+				return fmt.Errorf("contact sheet: %w", err)
+			}
+			fmt.Println(sheet)
+		} else {
+			fmt.Fprintln(os.Stderr, "imagen compare: all backends failed; no contact sheet written")
+		}
+	}
+	return nil
+}
+
+// compareResult is one backend's output in a comparison run. Error is set
+// when Generate failed for this backend; ImagePath + Metadata are empty in
+// that case.
+type compareResult struct {
+	Backend     string         `json:"backend"`
+	ImagePath   string         `json:"image_path,omitempty"`
+	Seed        int64          `json:"seed"`
+	LatencyMs   int64          `json:"latency_ms,omitempty"`
+	Model       string         `json:"model,omitempty"`
+	VRAMUsedMiB int64          `json:"vram_used_mib,omitempty"`
+	Metadata    map[string]any `json:"metadata,omitempty"`
+	Error       string         `json:"error,omitempty"`
+}
+
+func generateOne(ctx context.Context, cfg *config.Config, name, finalPrompt, negative string, w, h int, seed int64, steps int, runDir, rawPrompt string) (compareResult, error) {
+	be, err := buildBackend(cfg, name)
+	if err != nil {
+		return compareResult{Backend: name}, err
+	}
+	attachUsageSink(be)
+
+	req := backend.Request{
+		Prompt:         finalPrompt,
+		NegativePrompt: negative,
+		Width:          w,
+		Height:         h,
+		Steps:          steps,
+		Seed:           seed,
+	}
+	res, err := be.Generate(ctx, req)
+	if err != nil {
+		return compareResult{Backend: name}, err
+	}
+	defer res.ImageReader.Close()
+
+	imgBytes, err := io.ReadAll(res.ImageReader)
+	if err != nil {
+		return compareResult{Backend: name}, fmt.Errorf("read image: %w", err)
+	}
+
+	imgPath := filepath.Join(runDir, output.Slug(rawPrompt)+"--"+output.Slug(name)+"."+extFromMime(res.MimeType))
+	if err := os.WriteFile(imgPath, imgBytes, 0o644); err != nil {
+		return compareResult{Backend: name}, fmt.Errorf("write %s: %w", imgPath, err)
+	}
+
+	cr := compareResult{
+		Backend:   name,
+		ImagePath: imgPath,
+		Seed:      seedFromMetadata(res.Metadata, seed),
+		LatencyMs: metaInt64(res.Metadata, "latency_ms"),
+		Model:     metaString(res.Metadata, "model"),
+		Metadata:  res.Metadata,
+	}
+	if v, ok := res.Metadata["vram_used_mib"].(int64); ok {
+		cr.VRAMUsedMiB = v
+	}
+	return cr, nil
+}
+
+func successfulResults(rs []compareResult) []compareResult {
+	out := make([]compareResult, 0, len(rs))
+	for _, r := range rs {
+		if r.Error == "" && r.ImagePath != "" {
+			out = append(out, r)
+		}
+	}
+	return out
+}
+
+func writeCompareSidecar(path, rawPrompt, style, negative string, w, h int, seed int64, steps int, results []compareResult) error {
+	body := map[string]any{
+		"timestamp":  time.Now().UTC().Format(time.RFC3339),
+		"prompt":     rawPrompt,
+		"style":      style,
+		"negative":   negative,
+		"width":      w,
+		"height":     h,
+		"seed":       seed,
+		"steps":      steps,
+		"results":    results,
+		"backends":   backendNames(results),
+		"successful": len(successfulResults(results)),
+		"total":      len(results),
+	}
+	data, err := json.MarshalIndent(body, "", "  ")
+	if err != nil {
+		return fmt.Errorf("marshal sidecar: %w", err)
+	}
+	return os.WriteFile(path, append(data, '\n'), 0o644)
+}
+
+func backendNames(rs []compareResult) []string {
+	out := make([]string, len(rs))
+	for i, r := range rs {
+		out[i] = r.Backend
+	}
+	sort.Strings(out)
+	return out
+}
+
+// writeContactSheet stitches a grid of (image, label) cells into one PNG.
+// Cells are sized to fit in a target width of ~2400px while keeping each
+// individual image full-resolution (no downscale) up to the column limit;
+// past that, images sit at their native size and we just lay them out.
+//
+// The grid is a simple horizontal row when N <= 4; otherwise N/2 rows of 2.
+// This is a contact sheet, not a fancy gallery — readability for side-by-
+// side eyeballing is the goal.
+func writeContactSheet(path, prompt string, results []compareResult) error {
+	if len(results) == 0 {
+		return fmt.Errorf("no successful results to lay out")
+	}
+	cells := make([]contactCell, 0, len(results))
+	for _, r := range results {
+		img, err := readPNG(r.ImagePath)
+		if err != nil {
+			return fmt.Errorf("read %s: %w", r.ImagePath, err)
+		}
+		cells = append(cells, contactCell{
+			Image:    img,
+			Label:    r.Backend,
+			SubLabel: fmt.Sprintf("%dms · seed %d", r.LatencyMs, r.Seed),
+		})
+	}
+
+	cols := len(cells)
+	if cols > 4 {
+		cols = 2
+	}
+	rows := (len(cells) + cols - 1) / cols
+
+	const labelH = 64
+	const pad = 16
+
+	cellW := cells[0].Image.Bounds().Dx()
+	cellH := cells[0].Image.Bounds().Dy()
+	for _, c := range cells {
+		if w := c.Image.Bounds().Dx(); w > cellW {
+			cellW = w
+		}
+		if h := c.Image.Bounds().Dy(); h > cellH {
+			cellH = h
+		}
+	}
+
+	totalW := cols*cellW + (cols+1)*pad
+	totalH := rows*(cellH+labelH) + (rows+1)*pad + 48 // header band
+
+	canvas := image.NewRGBA(image.Rect(0, 0, totalW, totalH))
+	draw.Draw(canvas, canvas.Bounds(), &image.Uniform{C: color.RGBA{R: 30, G: 30, B: 35, A: 255}}, image.Point{}, draw.Src)
+
+	// Header: show the truncated prompt.
+	headerText := "imagen compare — " + truncate(prompt, 100)
+	drawText(canvas, headerText, pad, 30, color.RGBA{R: 240, G: 240, B: 245, A: 255})
+
+	for i, c := range cells {
+		col := i % cols
+		row := i / cols
+		x0 := pad + col*(cellW+pad)
+		y0 := 48 + pad + row*(cellH+labelH+pad)
+		// Center the image inside the cell when smaller than the max cell size.
+		iw := c.Image.Bounds().Dx()
+		ih := c.Image.Bounds().Dy()
+		offX := (cellW - iw) / 2
+		offY := (cellH - ih) / 2
+		dstRect := image.Rect(x0+offX, y0+offY, x0+offX+iw, y0+offY+ih)
+		draw.Draw(canvas, dstRect, c.Image, c.Image.Bounds().Min, draw.Src)
+
+		// Label band underneath.
+		labelY := y0 + cellH + 20
+		drawText(canvas, c.Label, x0+8, labelY, color.RGBA{R: 250, G: 250, B: 250, A: 255})
+		drawText(canvas, c.SubLabel, x0+8, labelY+22, color.RGBA{R: 180, G: 180, B: 190, A: 255})
+	}
+
+	f, err := os.Create(path)
+	if err != nil {
+		return fmt.Errorf("create %s: %w", path, err)
+	}
+	defer f.Close()
+	return png.Encode(f, canvas)
+}
+
+type contactCell struct {
+	Image    image.Image
+	Label    string
+	SubLabel string
+}
+
+func readPNG(path string) (image.Image, error) {
+	f, err := os.Open(path)
+	if err != nil {
+		return nil, err
+	}
+	defer f.Close()
+	img, _, err := image.Decode(f)
+	return img, err
+}
+
+func drawText(dst *image.RGBA, s string, x, y int, c color.Color) {
+	drawer := &font.Drawer{
+		Dst:  dst,
+		Src:  &image.Uniform{C: c},
+		Face: basicfont.Face7x13,
+		Dot:  fixed.Point26_6{X: fixed.I(x), Y: fixed.I(y)},
+	}
+	drawer.DrawString(s)
+}
+
+func truncate(s string, max int) string {
+	if len(s) <= max {
+		return s
+	}
+	return s[:max-1] + "…"
+}
+
+func splitCSV(s string) []string {
+	parts := strings.Split(s, ",")
+	out := make([]string, 0, len(parts))
+	for _, p := range parts {
+		p = strings.TrimSpace(p)
+		if p != "" {
+			out = append(out, p)
+		}
+	}
+	return out
+}
+
+func metaInt64(m map[string]any, key string) int64 {
+	v, ok := m[key]
+	if !ok {
+		return 0
+	}
+	switch n := v.(type) {
+	case int64:
+		return n
+	case int:
+		return int64(n)
+	case float64:
+		return int64(n)
+	}
+	return 0
+}

cmd/imagen/compare_test.go

@@ -0,0 +1,203 @@
+package main
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"image/png"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+)
+
+// runCompareWithEnv runs the compare subcommand in a writable tmpdir, with
+// XDG_CONFIG_HOME pointing somewhere empty so no host imagen.yaml leaks in.
+func runCompareWithEnv(t *testing.T, args []string) (stderr, stdout *bytes.Buffer, runDir string, err error) {
+	t.Helper()
+	tmp := t.TempDir()
+	t.Setenv("XDG_CONFIG_HOME", filepath.Join(tmp, "no-config"))
+	t.Setenv("HOME", tmp)
+
+	out := filepath.Join(tmp, "compare")
+	// stdlib flag parsing requires flags after the leading positional. Append
+	// --output at the end so any caller-supplied flags still parse cleanly.
+	args = append(args, "--output", out)
+
+	// Capture stdout/stderr via os pipes since runCompare writes directly.
+	oldStdout := os.Stdout
+	oldStderr := os.Stderr
+	rOut, wOut, _ := os.Pipe()
+	rErr, wErr, _ := os.Pipe()
+	os.Stdout = wOut
+	os.Stderr = wErr
+	defer func() {
+		os.Stdout = oldStdout
+		os.Stderr = oldStderr
+	}()
+
+	cmdErr := runCompare(context.Background(), args)
+
+	_ = wOut.Close()
+	_ = wErr.Close()
+	stdout = &bytes.Buffer{}
+	stderr = &bytes.Buffer{}
+	_, _ = stdout.ReadFrom(rOut)
+	_, _ = stderr.ReadFrom(rErr)
+
+	entries, _ := os.ReadDir(out)
+	if len(entries) == 1 {
+		runDir = filepath.Join(out, entries[0].Name())
+	}
+	return stderr, stdout, runDir, cmdErr
+}
+
+func TestCompareHappyPathWithMockBackends(t *testing.T) {
+	// Two mock instances stand in for two different backends. mock ignores
+	// cfg so we can reuse the registered type as the instance name and skip
+	// writing imagen.yaml entirely.
+	stderr, stdout, runDir, err := runCompareWithEnv(t, []string{
+		"a cat in a fishbowl",
+		"--models", "mock,mock",
+		"--size", "64x64",
+		"--seed", "42",
+	})
+	if err != nil {
+		t.Fatalf("runCompare: %v\nstderr: %s", err, stderr.String())
+	}
+
+	if runDir == "" {
+		t.Fatal("expected a run directory under --output")
+	}
+	// Sidecar JSON
+	sidecar := filepath.Join(runDir, "compare.json")
+	data, err := os.ReadFile(sidecar)
+	if err != nil {
+		t.Fatalf("read sidecar: %v", err)
+	}
+	var body struct {
+		Prompt     string `json:"prompt"`
+		Successful int    `json:"successful"`
+		Total      int    `json:"total"`
+		Results    []struct {
+			Backend   string `json:"backend"`
+			ImagePath string `json:"image_path"`
+			Error     string `json:"error"`
+		} `json:"results"`
+	}
+	if err := json.Unmarshal(data, &body); err != nil {
+		t.Fatalf("parse sidecar: %v\n%s", err, data)
+	}
+	if body.Prompt != "a cat in a fishbowl" {
+		t.Errorf("prompt = %q", body.Prompt)
+	}
+	if body.Total != 2 || body.Successful != 2 {
+		t.Errorf("counts = %d successful / %d total", body.Successful, body.Total)
+	}
+	for _, r := range body.Results {
+		if r.Error != "" {
+			t.Errorf("backend %s errored: %s", r.Backend, r.Error)
+		}
+		if _, err := os.Stat(r.ImagePath); err != nil {
+			t.Errorf("image not on disk for %s: %v", r.Backend, err)
+		}
+	}
+
+	// Contact sheet path was printed on stdout.
+	sheet := strings.TrimSpace(stdout.String())
+	if sheet == "" {
+		t.Fatal("expected contact sheet path on stdout")
+	}
+	f, err := os.Open(sheet)
+	if err != nil {
+		t.Fatalf("open contact sheet: %v", err)
+	}
+	defer f.Close()
+	img, err := png.Decode(f)
+	if err != nil {
+		t.Fatalf("decode contact sheet PNG: %v", err)
+	}
+	if w := img.Bounds().Dx(); w < 100 {
+		t.Errorf("contact sheet looks empty (width %d)", w)
+	}
+}
+
+func TestCompareSkipContactSheet(t *testing.T) {
+	stderr, stdout, runDir, err := runCompareWithEnv(t, []string{
+		"x",
+		"--models", "mock",
+		"--size", "32x32",
+		"--seed", "1",
+		"--no-contact-sheet",
+	})
+	if err != nil {
+		t.Fatalf("runCompare: %v\nstderr: %s", err, stderr.String())
+	}
+	if got := strings.TrimSpace(stdout.String()); got != "" {
+		t.Errorf("expected no stdout output (no contact sheet), got %q", got)
+	}
+	if _, err := os.Stat(filepath.Join(runDir, "contact-sheet.png")); err == nil {
+		t.Errorf("contact-sheet.png should not exist with --no-contact-sheet")
+	}
+}
+
+func TestCompareRecordsBackendErrors(t *testing.T) {
+	// One real (mock) + one unknown. Unknown should fail but not abort the
+	// run — sidecar records both, contact sheet built from successes only.
+	stderr, _, runDir, err := runCompareWithEnv(t, []string{
+		"y",
+		"--models", "mock,this-instance-does-not-exist",
+		"--size", "32x32",
+	})
+	if err != nil {
+		t.Fatalf("runCompare: %v\nstderr: %s", err, stderr.String())
+	}
+	sidecar := filepath.Join(runDir, "compare.json")
+	data, _ := os.ReadFile(sidecar)
+	var body struct {
+		Successful int `json:"successful"`
+		Total      int `json:"total"`
+		Results    []struct {
+			Backend string `json:"backend"`
+			Error   string `json:"error"`
+		} `json:"results"`
+	}
+	if err := json.Unmarshal(data, &body); err != nil {
+		t.Fatalf("parse sidecar: %v", err)
+	}
+	if body.Total != 2 {
+		t.Errorf("expected 2 results, got %d", body.Total)
+	}
+	if body.Successful != 1 {
+		t.Errorf("expected 1 success, got %d", body.Successful)
+	}
+	var sawError bool
+	for _, r := range body.Results {
+		if r.Backend == "this-instance-does-not-exist" && r.Error != "" {
+			sawError = true
+		}
+	}
+	if !sawError {
+		t.Errorf("expected an error recorded for the unknown backend")
+	}
+}
+
+func TestCompareNoModelsFails(t *testing.T) {
+	_, _, _, err := runCompareWithEnv(t, []string{"x"})
+	if err == nil {
+		t.Fatal("expected error when --models is empty")
+	}
+	if !strings.Contains(err.Error(), "--models") {
+		t.Errorf("error should mention --models, got %v", err)
+	}
+}
+
+func TestCompareNoPromptFails(t *testing.T) {
+	_, _, _, err := runCompareWithEnv(t, []string{"--models", "mock"})
+	if err == nil {
+		t.Fatal("expected error when prompt is missing")
+	}
+	if !strings.Contains(err.Error(), "missing prompt") {
+		t.Errorf("error should mention missing prompt, got %v", err)
+	}
+}

cmd/imagen/generate.go

@@ -132,9 +132,11 @@ func runGenerate(ctx context.Context, args []string) error {
 		fmt.Fprintln(os.Stderr, "sidecar:", paths.SidecarPath)
 	}
 
-	if err := maybeCloudSync(ctx, cfg, noCloud, paths, in, res, w, h); err != nil {
+	if result, err := maybeCloudSync(ctx, cfg, noCloud, "", "", paths, in, res, w, h); err != nil {
 		// cloud-sync failures are warnings — the image already wrote.
 		fmt.Fprintln(os.Stderr, "imagen: cloud sync:", err)
+	} else if result != nil && result.ImageID != "" {
+		fmt.Fprintf(os.Stderr, "cloud: imagen.images.id=%s storage_path=%s\n", result.ImageID, result.StoragePath)
 	}
 
 	if err := maybePreview(cfg, previewOn, previewOff, paths.ImagePath, rawPrompt); err != nil {
@@ -167,39 +169,48 @@ func resolveCloudSyncMode(cfg *config.Config, noCloudFlag bool, env string) (str
 }
 
 // maybeCloudSync resolves the effective mode and, if it says yes, uploads
-// the PNG and inserts the row. Always non-fatal — the image already wrote.
-func maybeCloudSync(ctx context.Context, cfg *config.Config, noCloud bool, paths *output.Outputs, in output.Inputs, res *backend.Result, width, height int) error {
+// the PNG and inserts the row. Returns the SyncResult on success so callers
+// that need the imagen.images.id (e.g. the worker linking a job row) can pick
+// it up. ownerOverride, when non-empty, wins over config + env — the worker
+// passes the job row's owner_user_id so each job is attributed correctly.
+// seriesID, when non-empty, lands on imagen.images.series_id so the
+// list-page query (`WHERE series_id IS NULL`) hides series members from
+// the flat grid; empty means solo run.
+func maybeCloudSync(ctx context.Context, cfg *config.Config, noCloud bool, ownerOverride, seriesID string, paths *output.Outputs, in output.Inputs, res *backend.Result, width, height int) (*cloud.SyncResult, error) {
 	mode, err := resolveCloudSyncMode(cfg, noCloud, os.Getenv("IMAGEN_CLOUD_SYNC"))
 	if err != nil {
-		return err
+		return nil, err
 	}
 	if mode == "off" {
-		return nil
+		return nil, nil
 	}
 
 	sink, ok := cloud.NewFromEnv()
 	if !ok {
 		if mode == "on" {
-			return fmt.Errorf("cloud_sync=on but SUPABASE_URL / SUPABASE_SERVICE_KEY not set in env")
+			return nil, fmt.Errorf("cloud_sync=on but SUPABASE_URL / SUPABASE_SERVICE_KEY not set in env")
 		}
 		// auto + missing env = silent skip.
-		return nil
+		return nil, nil
 	}
-	// Config-supplied owner_user_id takes precedence over $IMAGEN_OWNER_USER_ID.
-	if cfg != nil && cfg.OwnerUserID != "" {
+	switch {
+	case ownerOverride != "":
+		sink.OwnerUserID = ownerOverride
+	case cfg != nil && cfg.OwnerUserID != "":
+		// Config-supplied owner_user_id takes precedence over $IMAGEN_OWNER_USER_ID.
 		sink.OwnerUserID = cfg.OwnerUserID
 	}
 	if sink.OwnerUserID == "" {
 		if mode == "on" {
-			return fmt.Errorf("cloud_sync=on but owner_user_id not set in config and $IMAGEN_OWNER_USER_ID is empty")
+			return nil, fmt.Errorf("cloud_sync=on but owner_user_id not set in config and $IMAGEN_OWNER_USER_ID is empty")
 		}
 		// auto + missing UUID = silent skip.
-		return nil
+		return nil, nil
 	}
 
 	pngBytes, readErr := os.ReadFile(paths.ImagePath)
 	if readErr != nil {
-		return fmt.Errorf("read local image: %w", readErr)
+		return nil, fmt.Errorf("read local image: %w", readErr)
 	}
 
 	// Reuse the writer's date/slug/seed so storage_path mirrors the local
@@ -254,17 +265,11 @@ func maybeCloudSync(ctx context.Context, cfg *config.Config, noCloud bool, paths
 		LatencyMs:       latency,
 		CostUSDEstimate: cost,
 		Sidecar:         sidecar,
+		SeriesID:        seriesID,
 	}
 	syncCtx, cancel := context.WithTimeout(ctx, 60*time.Second)
 	defer cancel()
-	result, err := sink.Sync(syncCtx, syncReq)
-	if err != nil {
-		return err
-	}
-	if result != nil && result.ImageID != "" {
-		fmt.Fprintf(os.Stderr, "cloud: imagen.images.id=%s storage_path=%s\n", result.ImageID, result.StoragePath)
-	}
-	return nil
+	return sink.Sync(syncCtx, syncReq)
 }
 
 func metaString(m map[string]any, key string) string {

cmd/imagen/main.go

@@ -18,6 +18,9 @@ const helpText = `imagen — model-agnostic image generation
 
 Usage:
   imagen generate <prompt> [flags]    generate one image
+  imagen compare <prompt> --models a,b,c [flags]
+                                       run one prompt across N backends + contact sheet
+  imagen worker [flags]               consume the imagen.jobs queue (daemon)
   imagen backends                     list registered backend types
   imagen config init                  print a sample imagen.yaml on stdout
   imagen config validate              validate the active config
@@ -45,6 +48,10 @@ func main() {
 	switch os.Args[1] {
 	case "generate":
 		err = runGenerate(ctx, args)
+	case "compare":
+		err = runCompare(ctx, args)
+	case "worker":
+		err = runWorker(ctx, args)
 	case "backends":
 		err = runBackends(args)
 	case "config":

cmd/imagen/worker.go

@@ -0,0 +1,292 @@
+package main
+
+import (
+	"context"
+	"errors"
+	"flag"
+	"fmt"
+	"os"
+	"strings"
+	"time"
+
+	"github.com/jackc/pgx/v5"
+
+	"mgit.msbls.de/m/ImaGen/internal/backend"
+	"mgit.msbls.de/m/ImaGen/internal/config"
+	"mgit.msbls.de/m/ImaGen/internal/output"
+	"mgit.msbls.de/m/ImaGen/internal/prompt"
+	"mgit.msbls.de/m/ImaGen/internal/worker"
+)
+
+// runWorker is the `imagen worker` subcommand: a long-running daemon that
+// consumes the imagen.jobs queue and writes results into imagen.images via
+// the same cloud-sync path generate uses.
+func runWorker(ctx context.Context, args []string) error {
+	fs := flag.NewFlagSet("worker", flag.ContinueOnError)
+	var (
+		configPath   string
+		pollInterval time.Duration
+		jobTimeout   time.Duration
+	)
+	fs.StringVar(&configPath, "config", "", "config file path (default: ~/.config/imagen.yaml)")
+	fs.DurationVar(&pollInterval, "poll-interval", 5*time.Second, "safety-poll cadence between LISTEN wakeups")
+	fs.DurationVar(&jobTimeout, "job-timeout", 5*time.Minute, "max wall-time per job before the worker marks it failed")
+	fs.Usage = func() {
+		fmt.Fprintln(fs.Output(), `Usage: imagen worker [flags]
+
+Long-running daemon. LISTENs on the Postgres 'imagen_jobs' channel and polls
+imagen.jobs every --poll-interval as a safety net, claims pending rows, runs
+the generation pipeline, then updates the row with status + image_id.
+
+Env:
+  IMAGEN_WORKER_DATABASE_URL   Postgres DSN for direct LISTEN + UPDATE.
+                               Required (PostgREST cannot LISTEN).
+  SUPABASE_URL, SUPABASE_SERVICE_KEY, IMAGEN_OWNER_USER_ID
+                               Reused from generate's cloud-sync path; the
+                               worker writes imagen.images rows through the
+                               same code path. Per-job owner_user_id from the
+                               job row overrides IMAGEN_OWNER_USER_ID.`)
+		fs.PrintDefaults()
+	}
+	if err := fs.Parse(args); err != nil {
+		return err
+	}
+
+	cfg, cfgErr := config.Load(configPath)
+	if cfgErr != nil && !os.IsNotExist(cfgErr) {
+		return cfgErr
+	}
+
+	dsn := os.Getenv("IMAGEN_WORKER_DATABASE_URL")
+	if dsn == "" {
+		return userErr("IMAGEN_WORKER_DATABASE_URL not set; the worker needs a direct Postgres DSN for LISTEN/NOTIFY")
+	}
+
+	q, err := dialQueue(ctx, dsn)
+	if err != nil {
+		return fmt.Errorf("queue: %w", err)
+	}
+	defer q.Close()
+
+	p := &workerPipeline{cfg: cfg}
+	w := worker.New(q, p, worker.Config{
+		PollInterval: pollInterval,
+		JobTimeout:   jobTimeout,
+		Logger:       func(format string, a ...any) { fmt.Fprintf(os.Stderr, format+"\n", a...) },
+	})
+	fmt.Fprintln(os.Stderr, "imagen worker: ready (poll-interval", pollInterval, "job-timeout", jobTimeout, ")")
+	return w.Run(ctx)
+}
+
+// pgxQueue is the production Queue. It opens one dedicated connection used
+// for both LISTEN (long-lived) and UPDATE operations. A second connection
+// would split state needlessly — a single worker process processes one job
+// at a time so the connection is never contended.
+type pgxQueue struct {
+	conn *pgx.Conn
+}
+
+func dialQueue(ctx context.Context, dsn string) (*pgxQueue, error) {
+	conn, err := pgx.Connect(ctx, dsn)
+	if err != nil {
+		return nil, fmt.Errorf("pgx.Connect: %w", err)
+	}
+	if _, err := conn.Exec(ctx, "LISTEN imagen_jobs"); err != nil {
+		conn.Close(ctx)
+		return nil, fmt.Errorf("LISTEN imagen_jobs: %w", err)
+	}
+	return &pgxQueue{conn: conn}, nil
+}
+
+func (q *pgxQueue) Close() {
+	if q == nil || q.conn == nil {
+		return
+	}
+	// Best-effort: a 5s budget is enough to send a polite TerminateMessage.
+	shutdown, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+	defer cancel()
+	_ = q.conn.Close(shutdown)
+}
+
+// ClaimNextPending atomically marks the oldest pending row 'running' and
+// returns it. FOR UPDATE SKIP LOCKED is belt + braces against a second worker
+// process — out of scope for v1 but cheap insurance.
+func (q *pgxQueue) ClaimNextPending(ctx context.Context) (*worker.Job, error) {
+	// series_id is nullable on imagen.jobs (solo run when NULL); cast to text
+	// with COALESCE so pgx scans into a plain Go string. Empty string =
+	// solo run; the pipeline skips series propagation in that case.
+	const stmt = `
+		UPDATE imagen.jobs
+		SET status='running', started_at=now()
+		WHERE id = (
+			SELECT id FROM imagen.jobs
+			WHERE status='pending'
+			ORDER BY created_at
+			LIMIT 1
+			FOR UPDATE SKIP LOCKED
+		)
+		RETURNING id, owner_user_id, prompt, backend,
+		          COALESCE(model,''),
+		          COALESCE(width, 0), COALESCE(height, 0),
+		          COALESCE(steps, 0), COALESCE(seed, 0),
+		          COALESCE(style,''),
+		          COALESCE(series_id::text, '')`
+	var j worker.Job
+	err := q.conn.QueryRow(ctx, stmt).Scan(
+		&j.ID, &j.OwnerUserID, &j.Prompt, &j.Backend,
+		&j.Model, &j.Width, &j.Height, &j.Steps, &j.Seed, &j.Style,
+		&j.SeriesID,
+	)
+	if errors.Is(err, pgx.ErrNoRows) {
+		return nil, nil
+	}
+	if err != nil {
+		return nil, err
+	}
+	return &j, nil
+}
+
+func (q *pgxQueue) MarkDone(ctx context.Context, jobID, imageID string) error {
+	_, err := q.conn.Exec(ctx,
+		`UPDATE imagen.jobs SET status='done', image_id=$2, completed_at=now() WHERE id=$1`,
+		jobID, imageID)
+	return err
+}
+
+func (q *pgxQueue) MarkFailed(ctx context.Context, jobID, msg string) error {
+	// Trim outrageously long error text so a 10MB stack-trace doesn't end up
+	// in the row (callers see a summary, full text goes to stderr / logs).
+	const maxLen = 2000
+	if len(msg) > maxLen {
+		msg = msg[:maxLen] + "... [truncated]"
+	}
+	_, err := q.conn.Exec(ctx,
+		`UPDATE imagen.jobs SET status='failed', error=$2, completed_at=now() WHERE id=$1`,
+		jobID, msg)
+	return err
+}
+
+// WaitForJob blocks until a NOTIFY arrives on imagen_jobs, the timeout fires,
+// or ctx is cancelled. Notifications during a previous processJob are queued
+// by pgx and delivered on the next call — we don't lose wake-ups even when
+// processing took longer than poll-interval.
+func (q *pgxQueue) WaitForJob(ctx context.Context, timeout time.Duration) error {
+	waitCtx, cancel := context.WithTimeout(ctx, timeout)
+	defer cancel()
+	_, err := q.conn.WaitForNotification(waitCtx)
+	if err != nil {
+		if errors.Is(err, context.DeadlineExceeded) {
+			return nil // poll cadence fired
+		}
+		if errors.Is(err, context.Canceled) {
+			return context.Canceled
+		}
+		return err
+	}
+	return nil
+}
+
+// ResetStaleRunning bumps any rows stuck in 'running' back to 'pending' so
+// they get re-claimed. Called once at startup. A row stuck in 'running' came
+// from a previous worker crash; without this, flexsiebels would poll
+// forever on a job nobody is processing.
+func (q *pgxQueue) ResetStaleRunning(ctx context.Context) error {
+	_, err := q.conn.Exec(ctx,
+		`UPDATE imagen.jobs SET status='pending', started_at=NULL WHERE status='running'`)
+	return err
+}
+
+// workerPipeline is the Pipeline implementation that drives a single job
+// through buildBackend → prompt enrichment → generate → write disk →
+// cloud-sync, then returns the imagen.images.id back to the worker so it
+// can link the row.
+type workerPipeline struct {
+	cfg *config.Config
+}
+
+func (p *workerPipeline) Run(ctx context.Context, job worker.Job) worker.Outcome {
+	if job.OwnerUserID == "" {
+		return worker.Outcome{Err: fmt.Errorf("job %s: missing owner_user_id", job.ID)}
+	}
+	if job.Prompt == "" {
+		return worker.Outcome{Err: fmt.Errorf("job %s: empty prompt", job.ID)}
+	}
+	if job.Backend == "" {
+		return worker.Outcome{Err: fmt.Errorf("job %s: missing backend", job.ID)}
+	}
+
+	be, err := buildBackend(p.cfg, job.Backend)
+	if err != nil {
+		return worker.Outcome{Err: fmt.Errorf("backend %q: %w", job.Backend, err)}
+	}
+	attachUsageSink(be)
+
+	finalPrompt, err := prompt.Apply(job.Prompt, job.Style)
+	if err != nil {
+		return worker.Outcome{Err: fmt.Errorf("style: %w", err)}
+	}
+
+	req := backend.Request{
+		Prompt: finalPrompt,
+		Width:  job.Width,
+		Height: job.Height,
+		Steps:  job.Steps,
+		Seed:   job.Seed,
+		Style:  job.Style,
+	}
+	res, err := be.Generate(ctx, req)
+	if err != nil {
+		return worker.Outcome{Err: fmt.Errorf("generate: %w", err)}
+	}
+	defer res.ImageReader.Close()
+
+	writer := buildWriter(p.cfg, false)
+	in := output.Inputs{
+		Prompt:   job.Prompt,
+		Backend:  be.Name(),
+		Seed:     seedFromMetadata(res.Metadata, job.Seed),
+		Ext:      extFromMime(res.MimeType),
+		Metadata: res.Metadata,
+	}
+	paths, err := writer.Write(res.ImageReader, in)
+	if err != nil {
+		return worker.Outcome{Err: fmt.Errorf("write disk: %w", err)}
+	}
+
+	// Worker is queue-driven: cloud-sync is mandatory because flexsiebels
+	// needs imagen.images.id to render the result. Pass cloud_sync=on via
+	// the override path (third arg = ownerUserID); we set the mode by
+	// disallowing the 'off' branch through the cfg later if the user
+	// explicitly turned it off in config.
+	if cloudModeOff(p.cfg) {
+		// We refuse to silently drop a queued job. If cloud sync is off in
+		// config, the worker can't serve flexsiebels at all.
+		return worker.Outcome{Err: fmt.Errorf("output.cloud_sync=off in config; the worker requires cloud_sync=on or auto")}
+	}
+	syncRes, syncErr := maybeCloudSync(ctx, p.cfg, false, job.OwnerUserID, job.SeriesID, paths, in, res, dimOrFallback(job.Width, res, "width"), dimOrFallback(job.Height, res, "height"))
+	if syncErr != nil {
+		return worker.Outcome{Err: fmt.Errorf("cloud sync: %w", syncErr)}
+	}
+	if syncRes == nil || syncRes.ImageID == "" {
+		return worker.Outcome{Err: fmt.Errorf("cloud sync returned no imagen.images id (check SUPABASE_URL + SUPABASE_SERVICE_KEY)")}
+	}
+	return worker.Outcome{ImageID: syncRes.ImageID}
+}
+
+func cloudModeOff(cfg *config.Config) bool {
+	if cfg == nil {
+		return false
+	}
+	return strings.EqualFold(cfg.Output.CloudSync, "off")
+}
+
+// dimOrFallback returns job.<dim> when the job specified one, otherwise the
+// dimension reported by the backend's metadata. Some backends (Replicate
+// when given an aspect ratio) round the requested size to their nearest
+// supported value; this keeps the row honest about what was actually generated.
+func dimOrFallback(jobDim int, res *backend.Result, key string) int {
+	if jobDim > 0 {
+		return jobDim
+	}
+	return metaInt(res.Metadata, key)
+}

cmd/imagen/worker_integration_test.go

@@ -0,0 +1,129 @@
+package main
+
+import (
+	"context"
+	"fmt"
+	"os"
+	"testing"
+	"time"
+
+	"github.com/jackc/pgx/v5"
+
+	"mgit.msbls.de/m/ImaGen/internal/config"
+	"mgit.msbls.de/m/ImaGen/internal/worker"
+)
+
+// TestWorker_Integration_EndToEnd runs the full pipeline against a real
+// msupabase instance: insert a row into imagen.jobs, let the worker claim
+// it, generate via the mock backend (no Replicate spend, no ComfyUI
+// dependency), write to Supabase Storage + imagen.images, then flip the job
+// to 'done' with the linked image_id.
+//
+// Guarded by IMAGEN_WORKER_INTEGRATION=1. Required env beyond that:
+//
+//   IMAGEN_WORKER_DATABASE_URL  postgres DSN (direct, not PostgREST)
+//   SUPABASE_URL                e.g. https://supa.flexsiebels.de
+//   SUPABASE_SERVICE_KEY        service-role JWT
+//   IMAGEN_OWNER_USER_ID        UUID of an auth.users row (RLS fallback)
+//
+// The test creates and later deletes its own job row so repeated runs don't
+// leave debris.
+func TestWorker_Integration_EndToEnd(t *testing.T) {
+	if os.Getenv("IMAGEN_WORKER_INTEGRATION") != "1" {
+		t.Skip("set IMAGEN_WORKER_INTEGRATION=1 to run the integration test")
+	}
+	dsn := os.Getenv("IMAGEN_WORKER_DATABASE_URL")
+	if dsn == "" {
+		t.Fatal("IMAGEN_WORKER_DATABASE_URL must be set for the integration test")
+	}
+	if os.Getenv("SUPABASE_URL") == "" || os.Getenv("SUPABASE_SERVICE_KEY") == "" {
+		t.Fatal("SUPABASE_URL and SUPABASE_SERVICE_KEY must be set for the integration test")
+	}
+	owner := os.Getenv("IMAGEN_OWNER_USER_ID")
+	if owner == "" {
+		t.Fatal("IMAGEN_OWNER_USER_ID must be set for the integration test")
+	}
+
+	ctx, cancel := context.WithTimeout(context.Background(), 90*time.Second)
+	defer cancel()
+
+	q, err := dialQueue(ctx, dsn)
+	if err != nil {
+		t.Fatalf("dialQueue: %v", err)
+	}
+	defer q.Close()
+
+	// Insert the test job on a separate connection (the worker's conn is
+	// busy LISTENing). Mock backend = no external dependency.
+	insertConn, err := pgx.Connect(ctx, dsn)
+	if err != nil {
+		t.Fatalf("insert conn: %v", err)
+	}
+	defer insertConn.Close(ctx)
+
+	var jobID string
+	prompt := fmt.Sprintf("imagen integration test %d", time.Now().UnixNano())
+	err = insertConn.QueryRow(ctx, `
+		INSERT INTO imagen.jobs (owner_user_id, prompt, backend, width, height)
+		VALUES ($1, $2, 'mock', 64, 64)
+		RETURNING id`,
+		owner, prompt).Scan(&jobID)
+	if err != nil {
+		t.Fatalf("insert job: %v", err)
+	}
+	t.Logf("inserted imagen.jobs id=%s", jobID)
+	// Tidy up at the end of the test so a re-run starts clean.
+	defer func() {
+		cleanup, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+		defer cancel()
+		_, _ = insertConn.Exec(cleanup, `DELETE FROM imagen.jobs WHERE id=$1`, jobID)
+	}()
+
+	// Use a per-test temp dir so the generated PNG doesn't litter the repo.
+	tmpDir := t.TempDir()
+	cfg := &config.Config{Output: config.OutputConfig{Directory: tmpDir}}
+	p := &workerPipeline{cfg: cfg}
+	w := worker.New(q, p, worker.Config{
+		PollInterval: 1 * time.Second,
+		JobTimeout:   30 * time.Second,
+		Logger:       func(format string, a ...any) { t.Logf("worker: "+format, a...) },
+	})
+
+	// Run the worker until it processes one job (the one we just inserted)
+	// or the test context times out.
+	runCtx, runCancel := context.WithCancel(ctx)
+	done := make(chan struct{})
+	go func() {
+		_ = w.Run(runCtx)
+		close(done)
+	}()
+
+	// Poll for completion.
+	deadline := time.Now().Add(60 * time.Second)
+	var status, imageID string
+	for time.Now().Before(deadline) {
+		err = insertConn.QueryRow(ctx,
+			`SELECT status, COALESCE(image_id::text,'') FROM imagen.jobs WHERE id=$1`,
+			jobID).Scan(&status, &imageID)
+		if err != nil {
+			t.Fatalf("poll: %v", err)
+		}
+		if status == "done" || status == "failed" {
+			break
+		}
+		time.Sleep(500 * time.Millisecond)
+	}
+	runCancel()
+	<-done
+
+	if status != "done" {
+		var errText string
+		_ = insertConn.QueryRow(ctx,
+			`SELECT COALESCE(error,'') FROM imagen.jobs WHERE id=$1`, jobID).Scan(&errText)
+		t.Fatalf("job not done within timeout: status=%q error=%q", status, errText)
+	}
+	if imageID == "" {
+		t.Fatalf("job done but image_id is empty")
+	}
+	t.Logf("job done: image_id=%s", imageID)
+}

docs/architecture.md

@@ -7,7 +7,7 @@ upstream API. Each adapter only ever sees its own slice of `imagen.yaml`.
 
 ```
         ┌───────────────────────┐
-        │   cmd/imagen          │   CLI dispatch
+        │   cmd/imagen          │   CLI dispatch (generate / worker / …)
         │   (or HTTP server)    │
         └──────────┬────────────┘
                    │
@@ -18,6 +18,7 @@ upstream API. Each adapter only ever sees its own slice of `imagen.yaml`.
         │   internal/preview    │   tmux-img window spawner
         │   internal/cloud      │   Supabase Storage + imagen.images
         │   internal/usage      │   mai.imagen_usage cost-tracking
+        │   internal/worker     │   imagen.jobs queue consumer
         └──────────┬────────────┘
                    │
         ┌──────────▼────────────┐
@@ -105,9 +106,37 @@ contains the prompt, backend instance name, seed, ISO timestamp, and the
 - Network errors during `Generate` — wrap and return; no retry policy yet
   (decide per-adapter, or move to a shared retry helper if a pattern emerges).
 
+## Async write path: `imagen worker` + `imagen.jobs`
+
+`imagen generate` is the synchronous CLI. For web callers (flexsiebels'
+owner-mode UI) `cmd/imagen worker` runs as a daemon that consumes the
+`imagen.jobs` table.
+
+```
+flexsiebels POST          imagen worker (mRiver, systemd)
+  → INSERT INTO              LISTEN imagen_jobs  ◄── pg_notify trigger
+    imagen.jobs(pending)     claim row (UPDATE … RETURNING)
+                             dispatch through internal/backend
+                             write disk + cloud-sync via internal/cloud
+                             UPDATE imagen.jobs SET status='done', image_id=…
+```
+
+The queue table lives next to `imagen.images` in the same `imagen` schema.
+Owner-scoped RLS lets the flexsiebels user INSERT + read their own rows;
+the worker writes (status updates + image_id link) via service-role which
+bypasses RLS. A 5-second safety poll fires on every wake-up to cover
+dropped NOTIFY events and worker cold starts with a non-empty queue. See
+`docs/setup-worker-mriver.md` for the systemd installation.
+
+The worker reuses `internal/backend`, `internal/output`, and
+`internal/cloud` unchanged — it is purely an orchestration layer around
+the same pipeline `imagen generate` drives.
+
 ## Out of scope (today)
 
 - Image post-processing (cropping, watermarking).
-- Cost-tracking (lands with the Replicate adapter, since only API backends bill).
 - Multi-image `n>1` per request — backends that support it can expose it via
   `BackendOpts`; the framework doesn't have a first-class field yet.
+- Job cancellation / kill switch — separate follow-up issue.
+- Concurrent workers / multi-host scale-out — `FOR UPDATE SKIP LOCKED` in
+  the claim query makes it cheap to add, but a single worker is the v1 setup.

docs/backends.md

@@ -0,0 +1,310 @@
+# ImaGen backends
+
+This document covers the local-ComfyUI backend plug-in story: how adapters
+are layered, how to add a new model without touching Go, and the per-model
+setup steps for the bundled templates.
+
+For the host-side ComfyUI install (mRock — venv, weights for the default
+FLUX.1-schnell, systemd, VRAM coexistence with Ollama, smoke test against
+the raw HTTP API), see [`setup-comfyui-mrock.md`](setup-comfyui-mrock.md).
+
+## Architecture: Path 1 — workflow-template adapter
+
+`imagen generate` and `imagen compare` dispatch through the `comfyui`
+adapter, which holds the HTTP plumbing (`/prompt`, `/history/{id}`, `/view`,
+`/system_stats`) and treats the workflow itself as data. Each backend
+instance in `imagen.yaml` picks a workflow JSON via the `workflow:` key.
+Adding a new model is yaml + JSON, never Go:
+
+```
+internal/backend/
+  comfyui.go              # one adapter, all ComfyUI models
+  workflow_template.go    # loader + token-substitution
+  workflows/
+    flux1-schnell.json    # bundled templates (embedded with //go:embed)
+    flux2-klein.json
+    sd35-medium.json
+```
+
+### Why Path 1 over per-family adapters (`comfyui-flux.go`, `comfyui-sd3.go`…)
+
+- **Workflow JSON is the natural exchange format**. ComfyUI users export
+  workflows from its GUI as JSON. Anything else means rebuilding the graph
+  by hand in Go for every new model.
+- **Adding a model is a config change, not a build change**. With Path 2,
+  every new family is a Go file, a new test file, a registry entry, a new
+  worker binary, a redeploy. Path 1 lets us land a new model with one yaml
+  block + one JSON file + one section in this doc.
+- **The HTTP plumbing is identical across families**. `/prompt`,
+  `/history`, `/view`, the retry policy, the "value not in list" hint, VRAM
+  reporting — none of it depends on the workflow shape. Path 2 would
+  duplicate that across files.
+- **Failure isolation stays clean**. The workflow loader fails at adapter
+  construction (`imagen backends` surfaces the error), the HTTP layer
+  fails at `Generate`, and ComfyUI's own validation surfaces missing-model
+  hints. Each layer's error message points at the right config knob.
+
+Path 2's argument was "each family owns its quirks (samplers, schedulers,
+dual-stage etc.)". That argument doesn't survive contact with the
+substitution-map design: per-family knobs are just key/value fields in the
+yaml block and `${shift}`/`${guidance}`/`${cfg}` placeholders in the
+template. No code duplication, no inheritance to debug.
+
+### Token substitution
+
+`workflow_template.SubstituteWorkflow` walks the parsed JSON and replaces
+every whole-value string of the form `"${key}"` with the typed value from
+the substitution map. Numbers stay numbers, strings stay strings — no
+round-tripping through `strings.Replace`.
+
+The substitution map is built per call from:
+
+1. **Request fields** (always present): `${prompt}`, `${negative}`,
+   `${width}`, `${height}`, `${seed}`, `${steps}`, `${sampler}`,
+   `${scheduler}`, `${cfg}`.
+2. **Every scalar field from the yaml block** (string / int / int64 /
+   float64 / bool), minus framework keys (`type`, `base_url`, `workflow`,
+   `default_*`). So `${vae}`, `${clip}`, `${clip_l}`, `${clip_t5}`,
+   `${dtype}`, `${shift}`, `${guidance}` all become substitutable just by
+   being in yaml.
+3. **Sensible defaults** for the common optional knobs above, so a
+   workflow that references `${dtype}` without the user setting one in
+   yaml still substitutes cleanly (`fp8_e4m3fn` for FLUX, `3.0` for SD3
+   shift, etc.). Extra defaults are ignored by workflows that don't
+   reference them.
+
+Partial matches (e.g. `"prefix ${prompt} suffix"`) are deliberately **not**
+substituted — the placeholder must be the entire value so we can preserve
+its JSON type. This prevents a prompt containing literal `${seed}` text
+from corrupting the workflow.
+
+Unknown placeholders (referenced in JSON but missing from the substitution
+map) error out before the workflow leaves the binary.
+
+### Back-compat
+
+The `workflow:` field defaults to `flux1-schnell` if omitted. Existing
+yaml blocks like the pre-#10 FLUX.1-schnell instance:
+
+```yaml
+flux-schnell-local:
+  type: comfyui
+  base_url: http://mrock:8188
+  model: flux1-schnell.safetensors
+```
+
+still work unchanged — they implicitly pick up the migrated
+`flux1-schnell.json` template, which keeps the same node IDs (6, 8, 9, 10,
+11, 12, 13, 27, 30, 31) as the historical hardcoded workflow.
+
+## Bundled workflows
+
+### FLUX.1-schnell — the back-compat default
+
+| Field | Default | Notes |
+|---|---|---|
+| `model` | `flux1-schnell.safetensors` | drop in `models/unet/` |
+| `vae` | `ae.safetensors` | `models/vae/` |
+| `clip_l` | `clip_l.safetensors` | `models/clip/` |
+| `clip_t5` | `t5xxl_fp8_e4m3fn.safetensors` | `models/clip/` |
+| `dtype` | `fp8_e4m3fn` | weight dtype for the UNet loader |
+| `default_steps` / `default_cfg` | 4 / 1.0 | schnell is distilled to ~4 steps |
+
+VRAM peak ~10–12 GB at 1024×1024. Install path:
+[`setup-comfyui-mrock.md`](setup-comfyui-mrock.md). Already shipping.
+
+### FLUX.2 [klein] 4B — direct upgrade
+
+Released by Black Forest Labs late 2025 / early 2026, BFL non-commercial
+license. The distilled 4B "klein" variant lands sub-second on the RTX
+4070 Ti SUPER and shares the new Qwen-based text encoder + a re-trained
+VAE with the larger family.
+
+```yaml
+flux2-klein-local:
+  type: comfyui
+  base_url: http://mrock:8188
+  workflow: flux2-klein
+  model: flux-2-klein-base-4b-fp8.safetensors    # models/unet/
+  vae: flux2-vae.safetensors                     # models/vae/
+  clip: qwen_3_4b.safetensors                    # models/text_encoders/
+  dtype: fp8_e4m3fn
+  default_steps: 4
+  default_cfg: 1.0
+  guidance: 4.0
+```
+
+**Model downloads** (on mRock, ungated mirrors when available):
+
+```bash
+cd ~/dev/comfyui/models
+curl -L -o unet/flux-2-klein-base-4b-fp8.safetensors \
+  https://huggingface.co/black-forest-labs/FLUX.2-klein/resolve/main/flux-2-klein-base-4b-fp8.safetensors
+curl -L -o vae/flux2-vae.safetensors \
+  https://huggingface.co/black-forest-labs/FLUX.2-klein/resolve/main/flux2-vae.safetensors
+mkdir -p text_encoders
+curl -L -o text_encoders/qwen_3_4b.safetensors \
+  https://huggingface.co/black-forest-labs/FLUX.2-klein/resolve/main/qwen_3_4b.safetensors
+```
+
+BFL's primary repo is gated; if `curl` returns 401, configure an HF token
+in `~/.cache/huggingface/token` or use one of the community mirrors
+(check the official model card for the current list). The filenames the
+template references match BFL's canonical names — rename downloads to
+match if a mirror uses different ones.
+
+VRAM peak: ~8.5 GB (4B fp8). With Ollama parked at ~8 GB this still fits;
+unlike FLUX.1-schnell, klein doesn't require stopping Ollama on mRock.
+
+### SD3.5-medium — single-checkpoint variant
+
+Stability AI's 2.5B mid-size model with bundled text encoders. The
+`incl_clips_t5xxlfp8scaled` variant ships clip_g + clip_l + t5xxl_fp8 all
+in one `.safetensors`, so the workflow uses `CheckpointLoaderSimple`
+instead of separate UNet/VAE/CLIP loaders.
+
+```yaml
+sd35-medium-local:
+  type: comfyui
+  base_url: http://mrock:8188
+  workflow: sd35-medium
+  model: sd3.5_medium_incl_clips_t5xxlfp8scaled.safetensors  # models/checkpoints/
+  default_steps: 28
+  default_sampler: dpmpp_2m
+  default_scheduler: sgm_uniform
+  default_cfg: 4.5
+  shift: 3.0
+```
+
+**Model download** (on mRock):
+
+```bash
+cd ~/dev/comfyui/models
+curl -L -o checkpoints/sd3.5_medium_incl_clips_t5xxlfp8scaled.safetensors \
+  https://huggingface.co/stabilityai/stable-diffusion-3.5-medium/resolve/main/sd3.5_medium_incl_clips_t5xxlfp8scaled.safetensors
+```
+
+VRAM peak: ~9.9 GB at 1024×1024. Same envelope as FLUX.1-schnell — stop
+Ollama before generating, restart after.
+
+## Adding a new bundled workflow
+
+1. **Export from ComfyUI**: load the model in the ComfyUI GUI, build a
+   text-to-image workflow that produces what you want, "Save (API
+   Format)" — the file you get is the right shape.
+2. **Sprinkle placeholders**: open the JSON and replace per-call values
+   with `${name}` tokens. Whole-value substitution only:
+
+   ```json
+   "inputs": {
+     "text": "${prompt}",         // was "a cat sitting on a chair"
+     "seed": "${seed}",            // was 1234567
+     "steps": "${steps}",          // was 28
+     "cfg": "${cfg}",
+     "sampler_name": "${sampler}",
+     "scheduler": "${scheduler}",
+     "width": "${width}",
+     "height": "${height}"
+   }
+   ```
+
+   Use `${model}` for the checkpoint / unet filename and any per-template
+   knobs (`${vae}`, `${shift}`, `${guidance}`, `${clip}` …).
+3. **Drop it into `internal/backend/workflows/<name>.json`**. The
+   `//go:embed workflows/*.json` directive in `workflow_template.go`
+   picks it up at build time — no registry entry needed.
+4. **Add a yaml instance** in `internal/config/config.go`'s `Sample` block
+   for `imagen config init` (and `~/.config/imagen.yaml`) so users
+   discover the new backend.
+5. **Document the model files + HF download URLs** in this doc.
+6. **Smoke test**: `imagen generate "test" --backend <new-instance>
+   --size 1024x1024` should produce an image.
+
+Per-call overrides for sampler/scheduler/cfg go via `--steps`, `--seed`,
+and (programmatic) `backend.Request.BackendOpts["sampler"]` /
+`["scheduler"]` / `["cfg"]`. The compare harness forwards the
+constant-across-backends knobs verbatim.
+
+## Loading a workflow from disk (one-off)
+
+Pass an absolute filesystem path as `workflow:` and the adapter reads it
+from disk instead of the embedded FS. Handy for prototyping a new model
+before committing it:
+
+```yaml
+my-experimental:
+  type: comfyui
+  base_url: http://mrock:8188
+  workflow: /home/m/dev/comfyui/workflows/my-test.json
+  model: my-test-model.safetensors
+```
+
+The fallback chain is: filesystem path (if the string looks like a path
+or ends in `.json`), then bundled lookup by name, then bundled lookup
+with `.json` appended.
+
+## `imagen compare`: cross-backend evaluation
+
+```bash
+imagen compare "a wizard casting a spell" \
+  --models flux-schnell-local,flux2-klein-local,sd35-medium-local \
+  --size 1024x1024 \
+  --output ~/Pictures/imagen/compare
+```
+
+Per run, `compare`:
+
+- creates `<output>/<YYYYMMDD-HHMMSS>-<prompt-slug>/`
+- dispatches each named backend sequentially (mRock has one GPU; parallel
+  would OOM) — one backend's failure doesn't abort the run
+- writes per-backend PNGs as `<prompt-slug>--<backend-slug>.png`
+- writes `compare.json` listing every attempt (success + failure) with
+  per-model `seed`, `latency_ms`, `model`, `vram_used_mib`, full
+  `metadata` map, and the error string for any failure
+- composites a `contact-sheet.png` with the prompt as header and each
+  cell labelled `<backend>` / `<latency>ms · seed <n>`
+
+Flags mirror `generate`: `--seed`, `--steps`, `--style`, `--negative`,
+`--size` are shared across all backends. `--no-contact-sheet` skips the
+composite when only the per-image PNGs and sidecar matter (e.g. for a
+worker script that builds its own diff view).
+
+## Diagnostics
+
+`imagen backends` shows every instance with its registration state. For
+local ComfyUI, the status is currently just `registered` (we don't probe
+the upstream HTTP endpoint at startup — the boot-helper hint kicks in on
+first generation if mRock is asleep).
+
+Per-backend errors emit at most three kinds:
+
+1. **Adapter construction failure** (e.g. workflow JSON not found,
+   missing required yaml field). Caught at `buildBackend` time:
+   `imagen: backend "<name>": <err>`.
+2. **HTTP / runtime failure during Generate**. Wrapped with the boot
+   helper for `connection refused`/`no such host`/timeouts pointing at
+   `boot-whitetower mrock` so a sleeping mRock has an obvious next step.
+3. **ComfyUI workflow-validation failure** (200-with-node_errors or 400).
+   Surfaces with a model-not-found hint (matching `value_not_in_list` +
+   `unet_name`/`ckpt_name`) when applicable, pointing back at this doc.
+
+## Worker daemon notes
+
+`imagen worker` (the `imagen.jobs` queue consumer) uses the same adapter
++ workflow lookup as the synchronous CLI — flexsiebels' `/imagine` UI
+INSERTs a `backend = <instance>` row, the worker claims it, and the
+underlying ComfyUI HTTP calls are identical to what `generate` makes. No
+worker-specific changes are required when a new backend lands; the
+config + workflow are the only state that has to be present on the
+worker host.
+
+After merging a new template or yaml block:
+
+```bash
+# On the worker host (mRiver today):
+systemctl --user restart imagen-worker
+```
+
+The daemon-rebuild trap from issue #9 still applies: if you build the
+imagen binary on the dev machine and `scp` it over, restart the unit so
+systemd picks up the new ELF.

docs/setup-worker-mriver.md

@@ -0,0 +1,97 @@
+# `imagen worker` on mRiver
+
+The worker is a long-running daemon that consumes the `imagen.jobs` queue
+(written by flexsiebels' owner-mode UI) and writes the resulting image to
+Supabase Storage + `imagen.images` via the same cloud-sync path the CLI
+`imagen generate` uses.
+
+## Architecture
+
+```
+flexsiebels (owner UI)
+        |
+        v  INSERT INTO imagen.jobs (...)
+        |
+   msupabase Postgres
+        |
+        |   AFTER INSERT trigger:
+        |   pg_notify('imagen_jobs', NEW.id)
+        v
+   imagen worker (mRiver) ── LISTEN imagen_jobs
+        |
+        |   1. claim oldest 'pending' row (status='running')
+        |   2. dispatch to backend (FLUX schnell local / FLUX dev replicate / …)
+        |   3. write PNG to disk
+        |   4. upload to Storage + INSERT into imagen.images
+        |   5. UPDATE imagen.jobs SET status='done', image_id=...
+        v
+   flexsiebels polls GET .../jobs/<id> → renders the rendered card
+```
+
+A 5-second safety poll covers dropped NOTIFY events and worker cold starts
+with a non-empty queue.
+
+## One-time setup
+
+```bash
+# 1. Build the binary (or `task build`).
+cd ~/dev/ImaGen
+go build -o bin/imagen ./cmd/imagen
+
+# 2. Write the environment file.
+cp scripts/imagen-worker.env.example ~/.dotfiles/.env.imagen-worker
+chmod 600 ~/.dotfiles/.env.imagen-worker
+$EDITOR ~/.dotfiles/.env.imagen-worker   # fill in real DSN, service key
+
+# 3. Install the user systemd unit.
+mkdir -p ~/.config/systemd/user
+cp scripts/imagen-worker.service ~/.config/systemd/user/imagen-worker.service
+systemctl --user daemon-reload
+systemctl --user enable --now imagen-worker.service
+
+# 4. Tail the logs.
+journalctl --user -u imagen-worker -f
+```
+
+## Required env vars
+
+See `scripts/imagen-worker.env.example` for the canonical list. Required:
+
+- `IMAGEN_WORKER_DATABASE_URL` — direct Postgres DSN. PostgREST cannot LISTEN.
+- `SUPABASE_URL`, `SUPABASE_SERVICE_KEY` — same pair `imagen generate`
+  reads for the cloud-sync writer.
+- `IMAGEN_OWNER_USER_ID` — fallback owner UUID; per-job row's
+  `owner_user_id` overrides this.
+
+Optional, depending on enabled backends:
+
+- `REPLICATE_API_TOKEN` if any job will request a Replicate-typed backend.
+
+## Operating
+
+```bash
+systemctl --user status imagen-worker          # health
+systemctl --user restart imagen-worker         # pick up a new binary
+journalctl --user -u imagen-worker -n 200      # recent log lines
+```
+
+On startup the worker calls `ResetStaleRunning` once, flipping any rows
+left in `'running'` from a previous crash back to `'pending'` so they get
+re-claimed by the 5-second poll.
+
+## Smoke test
+
+With the worker running, INSERT a test job:
+
+```sql
+INSERT INTO imagen.jobs (owner_user_id, prompt, backend, width, height)
+VALUES (
+  'ac6c9501-3757-4a6d-8b97-2cff4288382b',
+  'a tiny owl wearing wire-rim glasses, photo',
+  'flux-schnell-local', 1024, 1024
+);
+```
+
+Within ~10 seconds the row should show `status='done'`, a populated
+`image_id` linking to a real `imagen.images` row, and a Storage object at
+`<YYYY-MM-DD>/<slug>-<seed>.png` in the `imagen-generated` bucket.

docs/usage.md

@@ -4,14 +4,21 @@
 
 ```
 imagen generate <prompt> [flags]    generate one image
+imagen compare <prompt> --models a,b,c [flags]
+                                    run one prompt across N backends + contact sheet
+imagen worker [flags]               consume the imagen.jobs queue (daemon)
 imagen backends                     list configured + registered backends
 imagen config init                  print a sample imagen.yaml on stdout
 imagen config validate              parse + validate the active config
 imagen config path                  print the resolved config path
 imagen serve [--addr :8080]         (stub) start the HTTP server
+imagen usage [--since DATE]         show cost-tracking rows
 imagen version                      print version
 ```
 
+For the per-backend setup (FLUX.1, FLUX.2 [klein], SD3.5 medium, …) and
+the architecture rationale, see [`backends.md`](backends.md).
+
 ## `generate` flags
 
 | Flag           | Default                      | Notes                                                       |

go.mod

@@ -1,5 +1,17 @@
 module mgit.msbls.de/m/ImaGen
 
-go 1.24
+go 1.25.0
 
-require gopkg.in/yaml.v3 v3.0.1
+require (
+	github.com/jackc/pgx/v5 v5.9.2
+	golang.org/x/image v0.40.0
+	gopkg.in/yaml.v3 v3.0.1
+)
+
+require (
+	github.com/jackc/pgpassfile v1.0.0 // indirect
+	github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761 // indirect
+	github.com/kr/text v0.2.0 // indirect
+	github.com/rogpeppe/go-internal v1.14.1 // indirect
+	golang.org/x/text v0.37.0 // indirect
+)

go.sum

@@ -1,4 +1,37 @@
-gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405 h1:yhCVgyC4o1eVCa2tZl7eS0r+SDo693bJlVdllGtEeKM=
+github.com/creack/pty v1.1.9/go.mod h1:oKZEueFk5CKHvIhNR5MUki03XCEU+Q6VDXinZuGJ33E=
+github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
+github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
+github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
+github.com/jackc/pgpassfile v1.0.0 h1:/6Hmqy13Ss2zCq62VdNG8tM1wchn8zjSGOBJ6icpsIM=
+github.com/jackc/pgpassfile v1.0.0/go.mod h1:CEx0iS5ambNFdcRtxPj5JhEz+xB6uRky5eyVu/W2HEg=
+github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761 h1:iCEnooe7UlwOQYpKFhBabPMi4aNAfoODPEFNiAnClxo=
+github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761/go.mod h1:5TJZWKEWniPve33vlWYSoGYefn3gLQRzjfDlhSJ9ZKM=
+github.com/jackc/pgx/v5 v5.9.2 h1:3ZhOzMWnR4yJ+RW1XImIPsD1aNSz4T4fyP7zlQb56hw=
+github.com/jackc/pgx/v5 v5.9.2/go.mod h1:mal1tBGAFfLHvZzaYh77YS/eC6IX9OWbRV1QIIM0Jn4=
+github.com/jackc/puddle/v2 v2.2.2 h1:PR8nw+E/1w0GLuRFSmiioY6UooMp6KJv0/61nB7icHo=
+github.com/jackc/puddle/v2 v2.2.2/go.mod h1:vriiEXHvEE654aYKXXjOvZM39qJ0q+azkZFrfEOc3H4=
+github.com/kr/pretty v0.3.0 h1:WgNl7dwNpEZ6jJ9k1snq4pZsg7DOEN8hP9Xw0Tsjwk0=
+github.com/kr/pretty v0.3.0/go.mod h1:640gp4NfQd8pI5XOwp5fnNeVWj67G7CFk/SaSQn7NBk=
+github.com/kr/text v0.2.0 h1:5Nx0Ya0ZqY2ygV366QzturHI13Jq95ApcVaJBhpS+AY=
+github.com/kr/text v0.2.0/go.mod h1:eLer722TekiGuMkidMxC/pM04lWEeraHUUmBw8l2grE=
+github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
+github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
+github.com/rogpeppe/go-internal v1.14.1 h1:UQB4HGPB6osV0SQTLymcB4TgvyWu6ZyliaW0tI/otEQ=
+github.com/rogpeppe/go-internal v1.14.1/go.mod h1:MaRKkUm5W0goXpeCfT7UZI6fk/L7L7so1lCWt35ZSgc=
+github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
+github.com/stretchr/testify v1.3.0/go.mod h1:M5WIy9Dh21IEIfnGCwXGc5bZfKNJtfHm1UVUgZn+9EI=
+github.com/stretchr/testify v1.7.0/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
+github.com/stretchr/testify v1.11.1 h1:7s2iGBzp5EwR7/aIZr8ao5+dra3wiQyKjjFuvgVKu7U=
+github.com/stretchr/testify v1.11.1/go.mod h1:wZwfW3scLgRK+23gO65QZefKpKQRnfz6sD981Nm4B6U=
+golang.org/x/image v0.40.0 h1:Tw4GyDXMo+daZN1znreBRC3VayR1aLFUyUEOLUdW1a8=
+golang.org/x/image v0.40.0/go.mod h1:uIc348UZMSvS5Z65CVZ7iDPaNobNFEPeJ4kbqTOszmA=
+golang.org/x/sync v0.20.0 h1:e0PTpb7pjO8GAtTs2dQ6jYa5BWYlMuX047Dco/pItO4=
+golang.org/x/sync v0.20.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
+golang.org/x/text v0.37.0 h1:Cqjiwd9eSg8e0QAkyCaQTNHFIIzWtidPahFWR83rTrc=
+golang.org/x/text v0.37.0/go.mod h1:a5sjxXGs9hsn/AJVwuElvCAo9v8QYLzvavO5z2PiM38=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
+gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c h1:Hei/4ADfdWqJk1ZMxUNpqntNwaWcugrBjAiHlqqRiVk=
+gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c/go.mod h1:JHkPIbrfpd72SG/EVd6muEfDQjcINNoR0C8j2r3qZ4Q=
+gopkg.in/yaml.v3 v3.0.0-20200313102051-9f266ea9e77c/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
 gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
 gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=

internal/backend/comfyui.go

@@ -20,24 +20,29 @@ import (
 const ComfyType = "comfyui"
 
 // Comfy is the ComfyUI adapter. It speaks the public `/prompt` + `/history`
-// + `/view` HTTP API and submits a fixed FLUX.1 schnell workflow built from
-// the values in Request.
+// + `/view` HTTP API and submits a workflow built by substituting Request
+// values into a JSON template (bundled under internal/backend/workflows/ or
+// loaded from a filesystem path).
 //
 // Concurrency: a single Comfy is safe to share across goroutines as long as
 // the underlying http.Client is. Generate does not hold long-lived state.
 type Comfy struct {
 	instance string
 
-	base    string
-	model   string
-	vae     string
-	clipL   string
-	clipT5  string
-	dtype   string
+	base     string
+	workflow string
+
+	// rawCfg keeps the original yaml block (minus framework keys) so we can
+	// expose every user-defined string/number as a workflow substitution
+	// without enumerating each per-model knob in Go. Empty values still get
+	// a substitution entry so a template can reference ${negative} when the
+	// request didn't pass one.
+	rawCfg map[string]any
 
 	defaultSteps     int
 	defaultSampler   string
 	defaultScheduler string
+	defaultCFG       float64
 
 	httpClient   *http.Client
 	pollInterval time.Duration
@@ -49,12 +54,20 @@ type Comfy struct {
 }
 
 // NewComfy is the registry constructor. cfg is the adapter's slice of
-// imagen.yaml. Required keys: base_url, model. The rest have sensible FLUX
-// schnell defaults.
+// imagen.yaml.
+//
+// Required keys: base_url, model.
+// Optional keys: workflow (defaults to "flux1-schnell" for back-compat with
+// existing configs), default_steps, default_sampler, default_scheduler,
+// default_cfg, plus any template-specific knobs (vae, clip, clip_l,
+// clip_t5, dtype, shift, guidance, …) the chosen workflow references.
 func NewComfy(name string, cfg map[string]any) (Backend, error) {
 	if name == "" {
 		return nil, fmt.Errorf("comfyui: empty instance name")
 	}
+	if cfg == nil {
+		cfg = map[string]any{}
+	}
 	base := strings.TrimRight(getString(cfg, "base_url", ""), "/")
 	if base == "" {
 		return nil, fmt.Errorf("comfyui[%s]: base_url is required", name)
@@ -67,23 +80,27 @@ func NewComfy(name string, cfg map[string]any) (Backend, error) {
 		return nil, fmt.Errorf("comfyui[%s]: model is required", name)
 	}
 
+	workflow := getString(cfg, "workflow", "flux1-schnell")
+	// Fail fast on a bad workflow ref so users see the error at startup,
+	// not on first /prompt submission.
+	if _, err := LoadWorkflowTemplate(workflow); err != nil {
+		return nil, fmt.Errorf("comfyui[%s]: %w", name, err)
+	}
+
 	c := &Comfy{
 		instance: name,
 		base:     base,
-		model:    model,
-
-		vae:    getString(cfg, "vae", "ae.safetensors"),
-		clipL:  getString(cfg, "clip_l", "clip_l.safetensors"),
-		clipT5: getString(cfg, "clip_t5", "t5xxl_fp8_e4m3fn.safetensors"),
-		dtype:  getString(cfg, "weight_dtype", "fp8_e4m3fn"),
+		workflow: workflow,
+		rawCfg:   cfg,
 
 		defaultSteps:     getInt(cfg, "default_steps", 4),
 		defaultSampler:   getString(cfg, "default_sampler", "euler"),
 		defaultScheduler: getString(cfg, "default_scheduler", "simple"),
+		defaultCFG:       getFloat(cfg, "default_cfg", 1.0),
 
 		httpClient:   &http.Client{Timeout: 60 * time.Second},
 		pollInterval: 250 * time.Millisecond,
-		pollTimeout:  120 * time.Second,
+		pollTimeout:  300 * time.Second,
 
 		randSeed:   cryptoSeed,
 		clientIDFn: randClientID,
@@ -103,19 +120,26 @@ func (c *Comfy) Generate(ctx context.Context, req Request) (*Result, error) {
 
 	sampler := c.defaultSampler
 	scheduler := c.defaultScheduler
+	cfg := c.defaultCFG
 	if v, ok := req.BackendOpts["sampler"].(string); ok && v != "" {
 		sampler = v
 	}
 	if v, ok := req.BackendOpts["scheduler"].(string); ok && v != "" {
 		scheduler = v
 	}
+	if v, ok := req.BackendOpts["cfg"].(float64); ok && v > 0 {
+		cfg = v
+	}
 
 	seed := req.Seed
 	if seed == 0 {
 		seed = c.randSeed()
 	}
 
-	workflow := c.buildWorkflow(req.Prompt, req.NegativePrompt, width, height, seed, steps, sampler, scheduler)
+	workflow, err := c.buildWorkflow(req.Prompt, req.NegativePrompt, width, height, seed, steps, sampler, scheduler, cfg)
+	if err != nil {
+		return nil, fmt.Errorf("comfyui[%s]: build workflow: %w", c.instance, err)
+	}
 	clientID := c.clientIDFn()
 
 	start := time.Now()
@@ -133,14 +157,17 @@ func (c *Comfy) Generate(ctx context.Context, req Request) (*Result, error) {
 	}
 	latencyMs := time.Since(start).Milliseconds()
 
+	model := getString(c.rawCfg, "model", "")
 	meta := map[string]any{
 		"backend":      c.instance,
 		"backend_type": ComfyType,
-		"model":        c.model,
+		"workflow":     c.workflow,
+		"model":        model,
 		"seed":         seed,
 		"steps":        steps,
 		"sampler":      sampler,
 		"scheduler":    scheduler,
+		"cfg":          cfg,
 		"width":        width,
 		"height":       height,
 		"latency_ms":   latencyMs,
@@ -173,6 +200,7 @@ func (c *Comfy) submitPrompt(ctx context.Context, workflow map[string]any, clien
 		return "", fmt.Errorf("comfyui: marshal workflow: %w", err)
 	}
 
+	model := getString(c.rawCfg, "model", "")
 	var lastErr error
 	for attempt := range 2 {
 		if attempt > 0 {
@@ -196,7 +224,7 @@ func (c *Comfy) submitPrompt(ctx context.Context, workflow map[string]any, clien
 		_ = resp.Body.Close()
 		switch {
 		case resp.StatusCode >= 200 && resp.StatusCode < 300:
-			return parsePromptID(respBody, c.model)
+			return parsePromptID(respBody, model)
 		case resp.StatusCode >= 500:
 			lastErr = fmt.Errorf("comfyui /prompt %d: %s", resp.StatusCode, snip(respBody))
 			continue
@@ -333,98 +361,74 @@ func (c *Comfy) connError(err error) error {
 // workflow-validation failures and put the diagnostics in node_errors; older
 // builds use 200 + node_errors. This handles the 4xx flavour.
 func (c *Comfy) classifyBadRequest(status int, body []byte) error {
-	if hint, ok := missingModelHint(body, c.model); ok {
-		return fmt.Errorf("comfyui /prompt %d: %s — see docs/setup-comfyui-mrock.md", status, hint)
+	model := getString(c.rawCfg, "model", "")
+	if hint, ok := missingModelHint(body, model); ok {
+		return fmt.Errorf("comfyui /prompt %d: %s — see docs/backends.md", status, hint)
 	}
 	return fmt.Errorf("comfyui /prompt %d: %s", status, snip(body))
 }
 
-// buildWorkflow assembles the canonical FLUX.1 schnell ComfyUI workflow,
-// node-IDs matching the upstream "flux-schnell" template so anyone debugging
-// in the ComfyUI UI sees a familiar shape.
-func (c *Comfy) buildWorkflow(prompt, negative string, w, h int, seed int64, steps int, sampler, scheduler string) map[string]any {
-	return map[string]any{
-		"6": map[string]any{
-			"class_type": "CLIPTextEncode",
-			"inputs": map[string]any{
-				"text": prompt,
-				"clip": []any{"11", 0},
-			},
-		},
-		"8": map[string]any{
-			"class_type": "VAEDecode",
-			"inputs": map[string]any{
-				"samples": []any{"31", 0},
-				"vae":     []any{"10", 0},
-			},
-		},
-		"9": map[string]any{
-			"class_type": "SaveImage",
-			"inputs": map[string]any{
-				"filename_prefix": "imagen",
-				"images":          []any{"8", 0},
-			},
-		},
-		"10": map[string]any{
-			"class_type": "VAELoader",
-			"inputs":     map[string]any{"vae_name": c.vae},
-		},
-		"11": map[string]any{
-			"class_type": "DualCLIPLoader",
-			"inputs": map[string]any{
-				"clip_name1": c.clipT5,
-				"clip_name2": c.clipL,
-				"type":       "flux",
-			},
-		},
-		"12": map[string]any{
-			"class_type": "UNETLoader",
-			"inputs": map[string]any{
-				"unet_name":    c.model,
-				"weight_dtype": c.dtype,
-			},
-		},
-		"13": map[string]any{
-			"class_type": "CLIPTextEncode",
-			"inputs": map[string]any{
-				"text": negative,
-				"clip": []any{"11", 0},
-			},
-		},
-		"27": map[string]any{
-			"class_type": "EmptySD3LatentImage",
-			"inputs": map[string]any{
-				"width":      w,
-				"height":     h,
-				"batch_size": 1,
-			},
-		},
-		"30": map[string]any{
-			"class_type": "ModelSamplingFlux",
-			"inputs": map[string]any{
-				"model":      []any{"12", 0},
-				"max_shift":  1.15,
-				"base_shift": 0.5,
-				"width":      w,
-				"height":     h,
-			},
-		},
-		"31": map[string]any{
-			"class_type": "KSampler",
-			"inputs": map[string]any{
-				"model":        []any{"30", 0},
-				"seed":         seed,
-				"steps":        steps,
-				"cfg":          1.0,
-				"sampler_name": sampler,
-				"scheduler":    scheduler,
-				"denoise":      1.0,
-				"positive":     []any{"6", 0},
-				"negative":     []any{"13", 0},
-				"latent_image": []any{"27", 0},
-			},
-		},
+// buildWorkflow loads the configured workflow template and substitutes the
+// per-call placeholders (prompt, seed, sampler, …) plus any string/number
+// fields the user defined in the yaml block. The set of placeholder keys
+// that aren't in `subs` produces an error from SubstituteWorkflow.
+func (c *Comfy) buildWorkflow(prompt, negative string, w, h int, seed int64, steps int, sampler, scheduler string, cfg float64) (map[string]any, error) {
+	wf, err := LoadWorkflowTemplate(c.workflow)
+	if err != nil {
+		return nil, err
 	}
+	subs := map[string]any{
+		"prompt":    prompt,
+		"negative":  negative,
+		"width":     w,
+		"height":    h,
+		"seed":      seed,
+		"steps":     steps,
+		"sampler":   sampler,
+		"scheduler": scheduler,
+		"cfg":       cfg,
+	}
+	// Surface every scalar field from the yaml block so per-template knobs
+	// (vae, clip, clip_l, clip_t5, dtype, shift, guidance, …) work without
+	// adapter-code changes. Framework keys are excluded.
+	for k, v := range c.rawCfg {
+		switch k {
+		case "type", "base_url", "workflow",
+			"default_steps", "default_sampler", "default_scheduler", "default_cfg":
+			continue
+		}
+		if _, alreadySet := subs[k]; alreadySet {
+			// A per-call var (e.g. ${prompt}) beats anything yaml put under
+			// the same key — yaml can't shadow request-derived values.
+			continue
+		}
+		switch v := v.(type) {
+		case string, int, int64, float64, bool:
+			subs[k] = v
+		}
+	}
+	// Provide sensible defaults for common optional knobs so a workflow that
+	// references one of these doesn't fail substitution when the user
+	// didn't override it in yaml. Extra keys are ignored if the workflow
+	// doesn't reference them, so it's safe to always set the lot.
+	defaults := map[string]any{
+		"vae":      "ae.safetensors",
+		"clip_l":   "clip_l.safetensors",
+		"clip_t5":  "t5xxl_fp8_e4m3fn.safetensors",
+		"clip":     "qwen_3_4b.safetensors",
+		"dtype":    "fp8_e4m3fn",
+		"guidance": 4.0,
+		"shift":    3.0,
+	}
+	for k, v := range defaults {
+		if _, ok := subs[k]; !ok {
+			subs[k] = v
+		}
+	}
+	if _, err := SubstituteWorkflow(wf, subs); err != nil {
+		return nil, err
+	}
+	return wf, nil
 }
 
 // parsePromptID handles the 2xx /prompt response. ComfyUI sometimes 200s a
@@ -432,8 +436,8 @@ func (c *Comfy) buildWorkflow(prompt, negative string, w, h int, seed int64, ste
 // turns that into the same user-facing error as a 4xx with the same body.
 func parsePromptID(body []byte, model string) (string, error) {
 	var resp struct {
-		PromptID   string         `json:"prompt_id"`
-		NodeErrors map[string]any `json:"node_errors"`
+		PromptID   string          `json:"prompt_id"`
+		NodeErrors map[string]any  `json:"node_errors"`
 		Error      json.RawMessage `json:"error"`
 	}
 	if err := json.Unmarshal(body, &resp); err != nil {
@@ -441,7 +445,7 @@ func parsePromptID(body []byte, model string) (string, error) {
 	}
 	if len(resp.NodeErrors) > 0 || len(resp.Error) > 0 {
 		if hint, ok := missingModelHint(body, model); ok {
-			return "", fmt.Errorf("comfyui /prompt: %s — see docs/setup-comfyui-mrock.md", hint)
+			return "", fmt.Errorf("comfyui /prompt: %s — see docs/backends.md", hint)
 		}
 		return "", fmt.Errorf("comfyui /prompt rejected workflow: %s", snip(body))
 	}
@@ -489,15 +493,21 @@ func parseHistory(body []byte, promptID string) (string, bool, error) {
 }
 
 // missingModelHint returns a user-actionable message when the response body
-// indicates the configured unet model isn't loaded on the server. ComfyUI
-// uses both the human-readable "Value not in list" message and the enum
-// "value_not_in_list" type — match either.
+// indicates the configured unet/checkpoint model isn't loaded on the server.
+// ComfyUI uses both the human-readable "Value not in list" message and the
+// enum "value_not_in_list" type — match either.
 func missingModelHint(body []byte, model string) (string, bool) {
 	s := string(body)
 	hasMarker := strings.Contains(s, "Value not in list") || strings.Contains(s, "value_not_in_list")
-	if hasMarker && strings.Contains(s, "unet_name") {
+	if !hasMarker {
+		return "", false
+	}
+	if strings.Contains(s, "unet_name") {
 		return fmt.Sprintf("model %q not present in the ComfyUI server's models/unet/", model), true
 	}
+	if strings.Contains(s, "ckpt_name") {
+		return fmt.Sprintf("checkpoint %q not present in the ComfyUI server's models/checkpoints/", model), true
+	}
 	return "", false
 }
 
@@ -536,6 +546,22 @@ func getInt(m map[string]any, k string, def int) int {
 	return def
 }
 
+func getFloat(m map[string]any, k string, def float64) float64 {
+	if v, ok := m[k]; ok {
+		switch n := v.(type) {
+		case float64:
+			return n
+		case float32:
+			return float64(n)
+		case int:
+			return float64(n)
+		case int64:
+			return float64(n)
+		}
+	}
+	return def
+}
+
 func orDefaultInt(v, def int) int {
 	if v == 0 {
 		return def

internal/backend/comfyui_test.go

@@ -312,7 +312,7 @@ func TestComfyMissingModelHintsAtSetupDoc(t *testing.T) {
 		t.Fatal("expected error")
 	}
 	msg := err.Error()
-	if !strings.Contains(msg, "docs/setup-comfyui-mrock.md") {
+	if !strings.Contains(msg, "docs/backends.md") {
 		t.Errorf("error should point at the setup doc, got %v", err)
 	}
 	if !strings.Contains(msg, "flux1-schnell.safetensors") {
@@ -331,7 +331,7 @@ func TestComfyMissingModelOn200WithNodeErrors(t *testing.T) {
 	if err == nil {
 		t.Fatal("expected error for node_errors on 200")
 	}
-	if !strings.Contains(err.Error(), "docs/setup-comfyui-mrock.md") {
+	if !strings.Contains(err.Error(), "docs/backends.md") {
 		t.Errorf("error should point at the setup doc, got %v", err)
 	}
 }

internal/backend/workflow_template.go

@@ -0,0 +1,156 @@
+package backend
+
+import (
+	"embed"
+	"encoding/json"
+	"fmt"
+	"io/fs"
+	"maps"
+	"os"
+	"path/filepath"
+	"regexp"
+	"sort"
+	"strings"
+)
+
+//go:embed workflows/*.json
+var bundledWorkflows embed.FS
+
+// placeholderRE matches a single-token placeholder like "${prompt}" — the
+// whole string value must be the placeholder, leading/trailing whitespace
+// allowed. This lets us preserve types (a numeric substitution becomes a
+// JSON number, not a stringified one) instead of round-tripping through
+// strings.Replace which would force everything into a string.
+var placeholderRE = regexp.MustCompile(`^\s*\$\{([a-zA-Z][a-zA-Z0-9_]*)\}\s*$`)
+
+// LoadWorkflowTemplate returns the parsed JSON for a workflow template.
+// `name` is resolved in this order:
+//
+//  1. exact filesystem path that exists on disk (absolute or relative);
+//  2. one of the bundled templates under internal/backend/workflows/
+//     (with or without the .json suffix).
+//
+// The returned map is a fresh deep copy of the template; callers can mutate
+// it freely.
+func LoadWorkflowTemplate(name string) (map[string]any, error) {
+	if name == "" {
+		return nil, fmt.Errorf("workflow template name is empty")
+	}
+	raw, err := readWorkflowBytes(name)
+	if err != nil {
+		return nil, err
+	}
+	var wf map[string]any
+	if err := json.Unmarshal(raw, &wf); err != nil {
+		return nil, fmt.Errorf("workflow %s: parse: %w", name, err)
+	}
+	return wf, nil
+}
+
+// BundledWorkflowNames returns the names of templates compiled into the
+// binary, sorted. Each name is the basename without the .json suffix.
+func BundledWorkflowNames() []string {
+	entries, err := fs.ReadDir(bundledWorkflows, "workflows")
+	if err != nil {
+		return nil
+	}
+	out := make([]string, 0, len(entries))
+	for _, e := range entries {
+		n := e.Name()
+		if !strings.HasSuffix(n, ".json") {
+			continue
+		}
+		out = append(out, strings.TrimSuffix(n, ".json"))
+	}
+	sort.Strings(out)
+	return out
+}
+
+func readWorkflowBytes(name string) ([]byte, error) {
+	// Filesystem path wins if it points at a real file. Lets a user override
+	// a bundled template by passing an absolute path in yaml.
+	if strings.ContainsRune(name, os.PathSeparator) || strings.HasSuffix(name, ".json") {
+		if b, err := os.ReadFile(name); err == nil {
+			return b, nil
+		} else if !os.IsNotExist(err) {
+			return nil, fmt.Errorf("workflow %s: %w", name, err)
+		}
+	}
+	// Bundled lookup. Try the literal name as a file inside workflows/, then
+	// with the .json suffix appended.
+	candidates := []string{
+		filepath.Join("workflows", name),
+		filepath.Join("workflows", name+".json"),
+	}
+	for _, c := range candidates {
+		if b, err := bundledWorkflows.ReadFile(c); err == nil {
+			return b, nil
+		}
+	}
+	return nil, fmt.Errorf("workflow %q not found (bundled templates: %v)", name, BundledWorkflowNames())
+}
+
+// SubstituteWorkflow walks wf and replaces every "${key}" string with the
+// matching value from subs, preserving JSON types. Returns the set of
+// placeholder keys it actually touched, so the caller can detect missing
+// substitutions even when a key is defined in subs but never referenced in
+// the workflow (typical when a yaml block sets a knob a different template
+// would consume).
+//
+// Unknown placeholders (referenced in the workflow but absent from subs)
+// produce an error so we never submit a workflow with raw "${foo}" tokens.
+func SubstituteWorkflow(wf map[string]any, subs map[string]any) (used map[string]struct{}, err error) {
+	used = make(map[string]struct{})
+	walked, err := substituteValue(wf, subs, used)
+	if err != nil {
+		return nil, err
+	}
+	// substituteValue returns the replacement for the top-level value, which
+	// should still be the same map (just with mutated children).
+	if m, ok := walked.(map[string]any); ok {
+		// Copy back into wf so the caller's reference reflects the result.
+		for k := range wf {
+			delete(wf, k)
+		}
+		maps.Copy(wf, m)
+	}
+	return used, nil
+}
+
+func substituteValue(v any, subs map[string]any, used map[string]struct{}) (any, error) {
+	switch x := v.(type) {
+	case map[string]any:
+		out := make(map[string]any, len(x))
+		for k, child := range x {
+			replaced, err := substituteValue(child, subs, used)
+			if err != nil {
+				return nil, err
+			}
+			out[k] = replaced
+		}
+		return out, nil
+	case []any:
+		out := make([]any, len(x))
+		for i, child := range x {
+			replaced, err := substituteValue(child, subs, used)
+			if err != nil {
+				return nil, err
+			}
+			out[i] = replaced
+		}
+		return out, nil
+	case string:
+		if m := placeholderRE.FindStringSubmatch(x); m != nil {
+			key := m[1]
+			val, ok := subs[key]
+			if !ok {
+				return nil, fmt.Errorf("workflow placeholder ${%s} has no substitution", key)
+			}
+			used[key] = struct{}{}
+			return val, nil
+		}
+		return x, nil
+	default:
+		return v, nil
+	}
+}

internal/backend/workflow_template_test.go

@@ -0,0 +1,153 @@
+package backend
+
+import (
+	"os"
+	"path/filepath"
+	"slices"
+	"strings"
+	"testing"
+)
+
+func TestBundledWorkflowsParseable(t *testing.T) {
+	names := BundledWorkflowNames()
+	if len(names) == 0 {
+		t.Fatal("expected at least one bundled workflow")
+	}
+	mustHave := []string{"flux1-schnell", "flux2-klein", "sd35-medium"}
+	for _, want := range mustHave {
+		if !slices.Contains(names, want) {
+			t.Errorf("bundled workflows missing %q (have: %v)", want, names)
+		}
+	}
+	// Every bundled template must parse and contain at least one node.
+	for _, n := range names {
+		wf, err := LoadWorkflowTemplate(n)
+		if err != nil {
+			t.Errorf("LoadWorkflowTemplate(%q): %v", n, err)
+			continue
+		}
+		if len(wf) == 0 {
+			t.Errorf("workflow %q has zero nodes", n)
+		}
+	}
+}
+
+func TestLoadWorkflowFromFilesystem(t *testing.T) {
+	dir := t.TempDir()
+	path := filepath.Join(dir, "custom.json")
+	body := `{"1":{"class_type":"X","inputs":{"v":"${prompt}"}}}`
+	if err := os.WriteFile(path, []byte(body), 0o644); err != nil {
+		t.Fatalf("write tmp workflow: %v", err)
+	}
+	wf, err := LoadWorkflowTemplate(path)
+	if err != nil {
+		t.Fatalf("load from path: %v", err)
+	}
+	if _, ok := wf["1"]; !ok {
+		t.Errorf("custom workflow missing node 1")
+	}
+}
+
+func TestLoadWorkflowUnknownNameErrors(t *testing.T) {
+	_, err := LoadWorkflowTemplate("definitely-not-a-real-workflow")
+	if err == nil {
+		t.Fatal("expected error for unknown workflow name")
+	}
+	if !strings.Contains(err.Error(), "not found") {
+		t.Errorf("error should say not found, got %v", err)
+	}
+}
+
+func TestSubstituteWorkflowPreservesTypes(t *testing.T) {
+	wf := map[string]any{
+		"31": map[string]any{
+			"class_type": "KSampler",
+			"inputs": map[string]any{
+				"seed":  "${seed}",
+				"steps": "${steps}",
+				"text":  "${prompt}",
+				"cfg":   "${cfg}",
+			},
+		},
+	}
+	subs := map[string]any{
+		"seed":   int64(42),
+		"steps":  11,
+		"prompt": "a cat",
+		"cfg":    4.5,
+	}
+	used, err := SubstituteWorkflow(wf, subs)
+	if err != nil {
+		t.Fatalf("Substitute: %v", err)
+	}
+	if len(used) != 4 {
+		t.Errorf("used = %v, want all four", used)
+	}
+	inputs := wf["31"].(map[string]any)["inputs"].(map[string]any)
+	if seed, ok := inputs["seed"].(int64); !ok || seed != 42 {
+		t.Errorf("seed = %T %v, want int64 42", inputs["seed"], inputs["seed"])
+	}
+	if steps, ok := inputs["steps"].(int); !ok || steps != 11 {
+		t.Errorf("steps = %T %v, want int 11", inputs["steps"], inputs["steps"])
+	}
+	if text, ok := inputs["text"].(string); !ok || text != "a cat" {
+		t.Errorf("text = %T %v, want string", inputs["text"], inputs["text"])
+	}
+	if cfg, ok := inputs["cfg"].(float64); !ok || cfg != 4.5 {
+		t.Errorf("cfg = %T %v, want float64 4.5", inputs["cfg"], inputs["cfg"])
+	}
+}
+
+func TestSubstituteWorkflowMissingPlaceholderErrors(t *testing.T) {
+	wf := map[string]any{
+		"1": map[string]any{"inputs": map[string]any{"v": "${missing}"}},
+	}
+	_, err := SubstituteWorkflow(wf, map[string]any{})
+	if err == nil {
+		t.Fatal("expected error for missing placeholder")
+	}
+	if !strings.Contains(err.Error(), "${missing}") {
+		t.Errorf("error should name the placeholder, got %v", err)
+	}
+}
+
+func TestSubstituteWorkflowOnlyWholeTokens(t *testing.T) {
+	// Partial-match strings ("prefix ${prompt} suffix") are NOT substituted —
+	// the placeholder must be the whole value so we can preserve types.
+	wf := map[string]any{
+		"1": map[string]any{"inputs": map[string]any{
+			"keep_string":  "stuff with ${prompt} inside",
+			"replace_full": "${prompt}",
+		}},
+	}
+	used, err := SubstituteWorkflow(wf, map[string]any{"prompt": "x"})
+	if err != nil {
+		t.Fatalf("Substitute: %v", err)
+	}
+	inputs := wf["1"].(map[string]any)["inputs"].(map[string]any)
+	if inputs["keep_string"].(string) != "stuff with ${prompt} inside" {
+		t.Errorf("partial match should be left alone, got %q", inputs["keep_string"])
+	}
+	if inputs["replace_full"].(string) != "x" {
+		t.Errorf("full-value match should substitute, got %q", inputs["replace_full"])
+	}
+	if _, ok := used["prompt"]; !ok {
+		t.Errorf("used should track keys that fired")
+	}
+}
+
+func TestFlux1SchnellTemplateMatchesLegacyShape(t *testing.T) {
+	// Regression guard against the historical hardcoded workflow: every
+	// node ID the old Comfy.buildWorkflow used must still exist in the
+	// migrated template.
+	wf, err := LoadWorkflowTemplate("flux1-schnell")
+	if err != nil {
+		t.Fatalf("load flux1-schnell: %v", err)
+	}
+	legacyNodes := []string{"6", "8", "9", "10", "11", "12", "13", "27", "30", "31"}
+	for _, id := range legacyNodes {
+		if _, ok := wf[id]; !ok {
+			t.Errorf("flux1-schnell template missing node %q (legacy parity)", id)
+		}
+	}
+}

internal/backend/workflows/flux1-schnell.json

@@ -0,0 +1,84 @@
+{
+  "6": {
+    "class_type": "CLIPTextEncode",
+    "inputs": {
+      "text": "${prompt}",
+      "clip": ["11", 0]
+    }
+  },
+  "8": {
+    "class_type": "VAEDecode",
+    "inputs": {
+      "samples": ["31", 0],
+      "vae": ["10", 0]
+    }
+  },
+  "9": {
+    "class_type": "SaveImage",
+    "inputs": {
+      "filename_prefix": "imagen",
+      "images": ["8", 0]
+    }
+  },
+  "10": {
+    "class_type": "VAELoader",
+    "inputs": {
+      "vae_name": "${vae}"
+    }
+  },
+  "11": {
+    "class_type": "DualCLIPLoader",
+    "inputs": {
+      "clip_name1": "${clip_t5}",
+      "clip_name2": "${clip_l}",
+      "type": "flux"
+    }
+  },
+  "12": {
+    "class_type": "UNETLoader",
+    "inputs": {
+      "unet_name": "${model}",
+      "weight_dtype": "${dtype}"
+    }
+  },
+  "13": {
+    "class_type": "CLIPTextEncode",
+    "inputs": {
+      "text": "${negative}",
+      "clip": ["11", 0]
+    }
+  },
+  "27": {
+    "class_type": "EmptySD3LatentImage",
+    "inputs": {
+      "width": "${width}",
+      "height": "${height}",
+      "batch_size": 1
+    }
+  },
+  "30": {
+    "class_type": "ModelSamplingFlux",
+    "inputs": {
+      "model": ["12", 0],
+      "max_shift": 1.15,
+      "base_shift": 0.5,
+      "width": "${width}",
+      "height": "${height}"
+    }
+  },
+  "31": {
+    "class_type": "KSampler",
+    "inputs": {
+      "model": ["30", 0],
+      "seed": "${seed}",
+      "steps": "${steps}",
+      "cfg": "${cfg}",
+      "sampler_name": "${sampler}",
+      "scheduler": "${scheduler}",
+      "denoise": 1.0,
+      "positive": ["6", 0],
+      "negative": ["13", 0],
+      "latent_image": ["27", 0]
+    }
+  }
+}

internal/backend/workflows/flux2-klein.json

@@ -0,0 +1,79 @@
+{
+  "6": {
+    "class_type": "CLIPTextEncode",
+    "inputs": {
+      "text": "${prompt}",
+      "clip": ["11", 0]
+    }
+  },
+  "8": {
+    "class_type": "VAEDecode",
+    "inputs": {
+      "samples": ["31", 0],
+      "vae": ["10", 0]
+    }
+  },
+  "9": {
+    "class_type": "SaveImage",
+    "inputs": {
+      "filename_prefix": "imagen",
+      "images": ["8", 0]
+    }
+  },
+  "10": {
+    "class_type": "VAELoader",
+    "inputs": {
+      "vae_name": "${vae}"
+    }
+  },
+  "11": {
+    "class_type": "CLIPLoader",
+    "inputs": {
+      "clip_name": "${clip}",
+      "type": "flux2"
+    }
+  },
+  "12": {
+    "class_type": "UNETLoader",
+    "inputs": {
+      "unet_name": "${model}",
+      "weight_dtype": "${dtype}"
+    }
+  },
+  "14": {
+    "class_type": "FluxGuidance",
+    "inputs": {
+      "conditioning": ["6", 0],
+      "guidance": "${guidance}"
+    }
+  },
+  "15": {
+    "class_type": "ConditioningZeroOut",
+    "inputs": {
+      "conditioning": ["6", 0]
+    }
+  },
+  "27": {
+    "class_type": "EmptyFlux2LatentImage",
+    "inputs": {
+      "width": "${width}",
+      "height": "${height}",
+      "batch_size": 1
+    }
+  },
+  "31": {
+    "class_type": "KSampler",
+    "inputs": {
+      "model": ["12", 0],
+      "seed": "${seed}",
+      "steps": "${steps}",
+      "cfg": "${cfg}",
+      "sampler_name": "${sampler}",
+      "scheduler": "${scheduler}",
+      "denoise": 1.0,
+      "positive": ["14", 0],
+      "negative": ["15", 0],
+      "latent_image": ["27", 0]
+    }
+  }
+}

internal/backend/workflows/sd35-medium.json

@@ -0,0 +1,66 @@
+{
+  "4": {
+    "class_type": "CheckpointLoaderSimple",
+    "inputs": {
+      "ckpt_name": "${model}"
+    }
+  },
+  "6": {
+    "class_type": "CLIPTextEncode",
+    "inputs": {
+      "text": "${prompt}",
+      "clip": ["4", 1]
+    }
+  },
+  "7": {
+    "class_type": "CLIPTextEncode",
+    "inputs": {
+      "text": "${negative}",
+      "clip": ["4", 1]
+    }
+  },
+  "8": {
+    "class_type": "VAEDecode",
+    "inputs": {
+      "samples": ["31", 0],
+      "vae": ["4", 2]
+    }
+  },
+  "9": {
+    "class_type": "SaveImage",
+    "inputs": {
+      "filename_prefix": "imagen",
+      "images": ["8", 0]
+    }
+  },
+  "13": {
+    "class_type": "ModelSamplingSD3",
+    "inputs": {
+      "model": ["4", 0],
+      "shift": "${shift}"
+    }
+  },
+  "27": {
+    "class_type": "EmptySD3LatentImage",
+    "inputs": {
+      "width": "${width}",
+      "height": "${height}",
+      "batch_size": 1
+    }
+  },
+  "31": {
+    "class_type": "KSampler",
+    "inputs": {
+      "model": ["13", 0],
+      "seed": "${seed}",
+      "steps": "${steps}",
+      "cfg": "${cfg}",
+      "sampler_name": "${sampler}",
+      "scheduler": "${scheduler}",
+      "denoise": 1.0,
+      "positive": ["6", 0],
+      "negative": ["7", 0],
+      "latent_image": ["27", 0]
+    }
+  }
+}

internal/cloud/cloud.go

@@ -100,6 +100,12 @@ type SyncRequest struct {
 	LatencyMs       int
 	CostUSDEstimate *float64
 	Sidecar         map[string]any
+
+	// SeriesID is the parent imagen.series row when this image is one of
+	// N tries in a batch. Empty means a solo run — the column stays NULL,
+	// which keeps the row visible on the main list-page query
+	// (`WHERE series_id IS NULL`).
+	SeriesID string
 }
 
 // SyncResult tells the caller what landed where.
@@ -195,6 +201,9 @@ func (s *Sink) insertRow(ctx context.Context, storagePath string, req SyncReques
 	if len(req.Sidecar) > 0 {
 		row["sidecar"] = req.Sidecar
 	}
+	if req.SeriesID != "" {
+		row["series_id"] = req.SeriesID
+	}
 
 	body, err := json.Marshal(row)
 	if err != nil {

internal/cloud/cloud_test.go

@@ -305,6 +305,54 @@ func TestSyncDBFailureSurfacesPathOnError(t *testing.T) {
 	}
 }
 
+// TestSyncWritesSeriesID is the second half of the ImaGen#9 propagation
+// contract: when SeriesID is non-empty, the POST body to imagen.images
+// carries `series_id`. When empty, the key is omitted entirely so the
+// row's series_id stays NULL (solo-run path, list-page query
+// `WHERE series_id IS NULL` keeps showing it).
+func TestSyncWritesSeriesID(t *testing.T) {
+	const seriesID = "22222222-2222-2222-2222-222222222222"
+	f := newFakeSupabase(t)
+	s := newSink(f.server)
+
+	_, err := s.Sync(context.Background(), SyncRequest{
+		Date: "2026-05-11", Slug: "x", Seed: 1, Ext: "png",
+		PNG: []byte("p"), Prompt: "p", Backend: "b",
+		SeriesID: seriesID,
+	})
+	if err != nil {
+		t.Fatalf("Sync: %v", err)
+	}
+	var row map[string]any
+	if err := json.Unmarshal(f.insertBody, &row); err != nil {
+		t.Fatalf("parse insert body: %v\n%s", err, f.insertBody)
+	}
+	if row["series_id"] != seriesID {
+		t.Fatalf("row.series_id = %v want %q", row["series_id"], seriesID)
+	}
+}
+
+func TestSyncOmitsSeriesIDWhenEmpty(t *testing.T) {
+	f := newFakeSupabase(t)
+	s := newSink(f.server)
+
+	_, err := s.Sync(context.Background(), SyncRequest{
+		Date: "2026-05-11", Slug: "x", Seed: 1, Ext: "png",
+		PNG: []byte("p"), Prompt: "p", Backend: "b",
+		// SeriesID intentionally empty.
+	})
+	if err != nil {
+		t.Fatalf("Sync: %v", err)
+	}
+	var row map[string]any
+	if err := json.Unmarshal(f.insertBody, &row); err != nil {
+		t.Fatalf("parse insert body: %v\n%s", err, f.insertBody)
+	}
+	if _, present := row["series_id"]; present {
+		t.Fatalf("solo run should omit series_id from POST body, got %v", row["series_id"])
+	}
+}
+
 func TestPathEscape(t *testing.T) {
 	cases := map[string]string{
 		"2026-05-11/lighthouse-42.png":              "2026-05-11/lighthouse-42.png",

internal/config/config.go

@@ -144,15 +144,54 @@ output:
   cloud_sync: auto
 
 backends:
+  # FLUX.1-schnell on the local ComfyUI server. The "workflow" key picks the
+  # bundled template under internal/backend/workflows/; omit it for back-compat
+  # (defaults to flux1-schnell). See docs/backends.md for the per-model setup.
   flux-schnell-local:
     type: comfyui
     base_url: http://mrock:8188
+    workflow: flux1-schnell
     # Filename of the unet checkpoint inside the ComfyUI server's
-    # models/unet/ directory. See docs/setup-comfyui-mrock.md.
+    # models/unet/ directory.
     model: flux1-schnell.safetensors
+    vae: ae.safetensors
+    clip_l: clip_l.safetensors
+    clip_t5: t5xxl_fp8_e4m3fn.safetensors
+    dtype: fp8_e4m3fn
     default_steps: 4
     default_sampler: euler
     default_scheduler: simple
+    default_cfg: 1.0
+
+  # FLUX.2 [klein] 4B distilled — sub-second on RTX 4070 Ti SUPER.
+  # Weights: BFL non-commercial; flux-2-klein-base-4b-fp8 in models/unet/,
+  # qwen_3_4b in models/text_encoders/, flux2-vae in models/vae/.
+  flux2-klein-local:
+    type: comfyui
+    base_url: http://mrock:8188
+    workflow: flux2-klein
+    model: flux-2-klein-base-4b-fp8.safetensors
+    vae: flux2-vae.safetensors
+    clip: qwen_3_4b.safetensors
+    dtype: fp8_e4m3fn
+    default_steps: 4
+    default_sampler: euler
+    default_scheduler: simple
+    default_cfg: 1.0
+    guidance: 4.0
+
+  # SD3.5 medium — single-checkpoint variant that bundles the three text
+  # encoders inside the .safetensors. Drop into models/checkpoints/.
+  sd35-medium-local:
+    type: comfyui
+    base_url: http://mrock:8188
+    workflow: sd35-medium
+    model: sd3.5_medium_incl_clips_t5xxlfp8scaled.safetensors
+    default_steps: 28
+    default_sampler: dpmpp_2m
+    default_scheduler: sgm_uniform
+    default_cfg: 4.5
+    shift: 3.0
 
   mock:
     type: mock

internal/prompt/prompt_test.go

@@ -37,7 +37,7 @@ func TestApplyToEmptyPromptUsesPresetOnly(t *testing.T) {
 }
 
 func TestStylesContainsAllExpected(t *testing.T) {
-	want := []string{"blog-header", "diagram", "illustration", "photo", "sketch"}
+	want := []string{"3d-render", "anime", "blog-header", "cinematic", "diagram", "illustration", "isometric", "line-art", "photo", "sketch", "watercolor"}
 	got := Styles()
 	if len(got) != len(want) {
 		t.Fatalf("Styles() = %v, want %v", got, want)

internal/prompt/styles.yaml

@@ -4,3 +4,9 @@ styles:
   diagram:      "minimal technical diagram, isometric, white background, line-art"
   sketch:       "rough pencil sketch, hand-drawn, monochrome"
   blog-header:  "wide aspect, conceptual, soft palette, editorial illustration"
+  cinematic:    "cinematic still, 35mm film, shallow depth of field, dramatic lighting, color graded"
+  watercolor:   "watercolor painting, soft washes, paper texture, loose brushwork"
+  anime:        "anime illustration, cel-shaded, expressive linework, vibrant flat colors"
+  3d-render:    "3d render, octane, soft global illumination, subtle ambient occlusion, physically based materials"
+  line-art:     "clean line art, black ink on white, no shading, even stroke weight"
+  isometric:    "isometric illustration, 30 degree projection, flat colors, crisp geometric shapes"

internal/worker/worker.go

@@ -0,0 +1,217 @@
+// Package worker consumes the imagen.jobs queue. It claims pending rows via
+// an UPDATE-returning lock (single source of truth, no double-claim window),
+// runs the supplied generation pipeline, then writes status + image_id back.
+//
+// The package is DB-agnostic: it talks to two small interfaces (Queue +
+// Pipeline) so unit tests can drive the claim/transition logic with no real
+// Postgres connection. cmd/imagen wires the pgx implementation.
+package worker
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"sync"
+	"time"
+)
+
+// Job is the slice of an imagen.jobs row the worker needs to drive a
+// generation. Null columns from the DB are represented as zero values; the
+// pipeline treats zero values as "use backend default" (same convention as
+// backend.Request).
+type Job struct {
+	ID          string
+	OwnerUserID string
+	Prompt      string
+	Backend     string
+	Model       string
+	Width       int
+	Height      int
+	Steps       int
+	Seed        int64
+	Style       string
+	// SeriesID is the parent imagen.series row when this job is one of N
+	// tries in a batch. Empty means a solo run — the pipeline must not
+	// propagate a series_id onto the resulting imagen.images row.
+	SeriesID string
+}
+
+// Outcome is what the pipeline reports back per job. ImageID is the
+// imagen.images.id the cloud-sync produced. Empty ImageID with nil Err means
+// the cloud-sync was skipped (config off) — we treat that as a failure for
+// the worker since flexsiebels needs the image_id to render the result.
+type Outcome struct {
+	ImageID string
+	Err     error
+}
+
+// Queue is the persistence layer for the imagen.jobs table. Implementations
+// must be safe for serialised single-worker use (concurrent claim across
+// multiple worker processes is out of scope for v1 — the FOR UPDATE SKIP
+// LOCKED clause in the pgx claim query covers it cheaply anyway).
+type Queue interface {
+	// ClaimNextPending atomically marks the oldest pending row 'running' and
+	// returns it. Returns (nil, nil) when the queue is empty.
+	ClaimNextPending(ctx context.Context) (*Job, error)
+	// MarkDone records success: status='done', image_id, completed_at=now().
+	MarkDone(ctx context.Context, jobID, imageID string) error
+	// MarkFailed records failure: status='failed', error=msg, completed_at=now().
+	MarkFailed(ctx context.Context, jobID, errMsg string) error
+	// WaitForJob blocks until either a NOTIFY arrives on imagen_jobs, the
+	// timeout expires, or ctx is cancelled. Returns nil on notification or
+	// timeout; returns ctx.Err() on cancellation. Transient connection errors
+	// are returned so the caller can decide to reconnect.
+	WaitForJob(ctx context.Context, timeout time.Duration) error
+	// ResetStaleRunning marks any rows stuck in 'running' (e.g. left over
+	// from a crash before this process started) back to 'pending'. Called
+	// once at worker startup so the cold-start safety poll can pick them up.
+	ResetStaleRunning(ctx context.Context) error
+}
+
+// Pipeline runs one generation and reports back the imagen.images.id (or an
+// error). The implementation owns backend dispatch, prompt enrichment, disk
+// write, and cloud-sync; the worker only orchestrates queue state.
+type Pipeline interface {
+	Run(ctx context.Context, job Job) Outcome
+}
+
+// Config is the runtime knob set for the worker loop.
+type Config struct {
+	// PollInterval is the safety-poll cadence between LISTEN wakeups. Picking
+	// this too low wastes DB roundtrips; too high lets a dropped NOTIFY
+	// stall the queue. 5s is the spec'd default.
+	PollInterval time.Duration
+	// JobTimeout caps any single Pipeline.Run. A backend hang shouldn't
+	// freeze the queue forever.
+	JobTimeout time.Duration
+	// Logger receives one-line status events. nil means silent.
+	Logger func(format string, args ...any)
+}
+
+// Worker is the orchestration loop. It is not reusable across Run calls.
+type Worker struct {
+	q      Queue
+	p      Pipeline
+	cfg    Config
+
+	// processingMu guards the in-flight job so SIGTERM-triggered shutdown
+	// waits for it to complete before returning.
+	processingMu sync.Mutex
+}
+
+// New constructs a Worker.
+func New(q Queue, p Pipeline, cfg Config) *Worker {
+	if cfg.PollInterval <= 0 {
+		cfg.PollInterval = 5 * time.Second
+	}
+	if cfg.JobTimeout <= 0 {
+		cfg.JobTimeout = 5 * time.Minute
+	}
+	return &Worker{q: q, p: p, cfg: cfg}
+}
+
+// Run drives the consume loop until ctx is cancelled or a fatal queue error
+// (e.g. unrecoverable DB drop) is returned. A LISTEN wait can fail with a
+// transient transport error; the worker logs and continues so a temporary
+// network blip doesn't take it down.
+func (w *Worker) Run(ctx context.Context) error {
+	if err := w.q.ResetStaleRunning(ctx); err != nil {
+		w.log("worker: reset stale running rows: %v", err)
+		// Don't return — a stale row will eventually be visible to the poll
+		// path once flexsiebels gives up and resubmits, and we'd rather keep
+		// serving fresh jobs than crash here.
+	}
+	for {
+		if err := ctx.Err(); err != nil {
+			return nil
+		}
+		// Drain the queue: claim and process until empty.
+		if err := w.drain(ctx); err != nil && !errors.Is(err, context.Canceled) {
+			w.log("worker: drain: %v", err)
+		}
+		if err := ctx.Err(); err != nil {
+			return nil
+		}
+		// Wait for the next wake. WaitForJob covers both LISTEN and the
+		// timeout-based poll fallback; either returns nil and we loop.
+		if err := w.q.WaitForJob(ctx, w.cfg.PollInterval); err != nil {
+			if errors.Is(err, context.Canceled) {
+				return nil
+			}
+			w.log("worker: wait: %v (continuing)", err)
+			// Pace the retries so a totally-broken DB doesn't busy-spin.
+			select {
+			case <-ctx.Done():
+				return nil
+			case <-time.After(w.cfg.PollInterval):
+			}
+		}
+	}
+}
+
+// drain claims and processes every currently-pending job. The job-scoped
+// context is derived from context.Background() so that a SIGTERM mid-job
+// still lets the pipeline finish — that's the "no half-state on shutdown"
+// guarantee the issue calls for.
+func (w *Worker) drain(ctx context.Context) error {
+	for {
+		if err := ctx.Err(); err != nil {
+			return err
+		}
+		job, err := w.q.ClaimNextPending(ctx)
+		if err != nil {
+			return fmt.Errorf("claim: %w", err)
+		}
+		if job == nil {
+			return nil
+		}
+		w.processOne(*job)
+	}
+}
+
+// processOne runs the pipeline for one already-claimed job and writes the
+// outcome back to the queue. The job context is independent of the outer
+// ctx so an in-flight job can finish even after SIGTERM.
+func (w *Worker) processOne(job Job) {
+	w.processingMu.Lock()
+	defer w.processingMu.Unlock()
+
+	w.log("worker: processing job %s backend=%s", job.ID, job.Backend)
+	jobCtx, cancel := context.WithTimeout(context.Background(), w.cfg.JobTimeout)
+	defer cancel()
+	out := w.p.Run(jobCtx, job)
+
+	// Status-update uses Background ctx with a short timeout — we must
+	// always be able to record the outcome, otherwise the row sits in
+	// 'running' forever.
+	updCtx, updCancel := context.WithTimeout(context.Background(), 30*time.Second)
+	defer updCancel()
+	if out.Err != nil {
+		w.log("worker: job %s failed: %v", job.ID, out.Err)
+		if err := w.q.MarkFailed(updCtx, job.ID, out.Err.Error()); err != nil {
+			w.log("worker: mark failed for %s: %v", job.ID, err)
+		}
+		return
+	}
+	if out.ImageID == "" {
+		// Pipeline reported success but no imagen.images row — treat as
+		// failure because flexsiebels has nothing to link.
+		const msg = "pipeline did not return an imagen.images id (cloud sync misconfigured?)"
+		w.log("worker: job %s: %s", job.ID, msg)
+		if err := w.q.MarkFailed(updCtx, job.ID, msg); err != nil {
+			w.log("worker: mark failed for %s: %v", job.ID, err)
+		}
+		return
+	}
+	if err := w.q.MarkDone(updCtx, job.ID, out.ImageID); err != nil {
+		w.log("worker: mark done for %s: %v", job.ID, err)
+		return
+	}
+	w.log("worker: job %s done image_id=%s", job.ID, out.ImageID)
+}
+
+func (w *Worker) log(format string, args ...any) {
+	if w.cfg.Logger != nil {
+		w.cfg.Logger(format, args...)
+	}
+}

internal/worker/worker_test.go

@@ -0,0 +1,376 @@
+package worker
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"sync"
+	"testing"
+	"time"
+)
+
+// fakeQueue is a hand-rolled in-memory queue that mirrors the contract of a
+// real Postgres-backed implementation: ClaimNextPending atomically takes one
+// pending row and flips its status to "running", MarkDone/MarkFailed are
+// idempotent terminal transitions, WaitForJob blocks until notified or until
+// the timeout elapses.
+type fakeQueue struct {
+	mu      sync.Mutex
+	pending []Job
+	state   map[string]string // jobID -> status
+	last    map[string]string // jobID -> error msg or image_id
+	notify  chan struct{}
+
+	claimErr error
+	doneErr  error
+	failErr  error
+	resetErr error
+
+	claimed int
+	done    int
+	failed  int
+	resets  int
+}
+
+func newFakeQueue(jobs ...Job) *fakeQueue {
+	q := &fakeQueue{
+		state:   make(map[string]string),
+		last:    make(map[string]string),
+		notify:  make(chan struct{}, 16),
+	}
+	for _, j := range jobs {
+		q.pending = append(q.pending, j)
+		q.state[j.ID] = "pending"
+	}
+	return q
+}
+
+func (q *fakeQueue) ClaimNextPending(ctx context.Context) (*Job, error) {
+	q.mu.Lock()
+	defer q.mu.Unlock()
+	if q.claimErr != nil {
+		return nil, q.claimErr
+	}
+	if len(q.pending) == 0 {
+		return nil, nil
+	}
+	j := q.pending[0]
+	q.pending = q.pending[1:]
+	q.state[j.ID] = "running"
+	q.claimed++
+	return &j, nil
+}
+
+func (q *fakeQueue) MarkDone(ctx context.Context, jobID, imageID string) error {
+	q.mu.Lock()
+	defer q.mu.Unlock()
+	if q.doneErr != nil {
+		return q.doneErr
+	}
+	q.state[jobID] = "done"
+	q.last[jobID] = imageID
+	q.done++
+	return nil
+}
+
+func (q *fakeQueue) MarkFailed(ctx context.Context, jobID, msg string) error {
+	q.mu.Lock()
+	defer q.mu.Unlock()
+	if q.failErr != nil {
+		return q.failErr
+	}
+	q.state[jobID] = "failed"
+	q.last[jobID] = msg
+	q.failed++
+	return nil
+}
+
+func (q *fakeQueue) WaitForJob(ctx context.Context, timeout time.Duration) error {
+	select {
+	case <-ctx.Done():
+		return ctx.Err()
+	case <-q.notify:
+		return nil
+	case <-time.After(timeout):
+		return nil
+	}
+}
+
+func (q *fakeQueue) ResetStaleRunning(ctx context.Context) error {
+	q.mu.Lock()
+	defer q.mu.Unlock()
+	q.resets++
+	return q.resetErr
+}
+
+// pingNotify simulates an INSERT-trigger NOTIFY by waking WaitForJob.
+func (q *fakeQueue) pingNotify() {
+	select {
+	case q.notify <- struct{}{}:
+	default:
+	}
+}
+
+// stub pipeline.
+type fakePipeline struct {
+	mu      sync.Mutex
+	results map[string]Outcome // by job.ID; "" key = default outcome
+	calls   int
+	delay   time.Duration
+	lastJob Job
+}
+
+func (p *fakePipeline) Run(ctx context.Context, job Job) Outcome {
+	p.mu.Lock()
+	p.calls++
+	p.lastJob = job
+	delay := p.delay
+	out, ok := p.results[job.ID]
+	if !ok {
+		out = p.results[""]
+	}
+	p.mu.Unlock()
+	if delay > 0 {
+		select {
+		case <-ctx.Done():
+			return Outcome{Err: ctx.Err()}
+		case <-time.After(delay):
+		}
+	}
+	return out
+}
+
+func TestWorker_DonePath(t *testing.T) {
+	q := newFakeQueue(
+		Job{ID: "j1", Prompt: "a", Backend: "mock"},
+	)
+	p := &fakePipeline{results: map[string]Outcome{"j1": {ImageID: "img-1"}}}
+	w := New(q, p, Config{PollInterval: 10 * time.Millisecond, JobTimeout: time.Second})
+	ctx, cancel := context.WithCancel(context.Background())
+	go func() {
+		time.Sleep(80 * time.Millisecond)
+		cancel()
+	}()
+	if err := w.Run(ctx); err != nil {
+		t.Fatalf("Run: %v", err)
+	}
+	if got := q.state["j1"]; got != "done" {
+		t.Fatalf("state=%q want done", got)
+	}
+	if got := q.last["j1"]; got != "img-1" {
+		t.Fatalf("image_id=%q want img-1", got)
+	}
+	if q.done != 1 || q.failed != 0 {
+		t.Fatalf("counts: done=%d failed=%d", q.done, q.failed)
+	}
+	if p.calls != 1 {
+		t.Fatalf("pipeline calls=%d want 1", p.calls)
+	}
+	if q.resets != 1 {
+		t.Fatalf("ResetStaleRunning calls=%d want 1", q.resets)
+	}
+}
+
+func TestWorker_FailedPath_RecordsErrorText(t *testing.T) {
+	q := newFakeQueue(Job{ID: "j1", Prompt: "a", Backend: "mock"})
+	p := &fakePipeline{results: map[string]Outcome{"j1": {Err: errors.New("backend unreachable")}}}
+	w := New(q, p, Config{PollInterval: 10 * time.Millisecond, JobTimeout: time.Second})
+	ctx, cancel := context.WithCancel(context.Background())
+	go func() { time.Sleep(80 * time.Millisecond); cancel() }()
+	_ = w.Run(ctx)
+
+	if got := q.state["j1"]; got != "failed" {
+		t.Fatalf("state=%q want failed", got)
+	}
+	if got := q.last["j1"]; got != "backend unreachable" {
+		t.Fatalf("error=%q want %q", got, "backend unreachable")
+	}
+	if q.done != 0 || q.failed != 1 {
+		t.Fatalf("counts: done=%d failed=%d", q.done, q.failed)
+	}
+}
+
+func TestWorker_MissingImageID_TreatedAsFailure(t *testing.T) {
+	q := newFakeQueue(Job{ID: "j1", Prompt: "a", Backend: "mock"})
+	// Outcome has neither Err nor ImageID — pipeline silently swallowed
+	// cloud-sync. flexsiebels needs the image_id; without it, fail the job.
+	p := &fakePipeline{results: map[string]Outcome{"j1": {}}}
+	w := New(q, p, Config{PollInterval: 10 * time.Millisecond, JobTimeout: time.Second})
+	ctx, cancel := context.WithCancel(context.Background())
+	go func() { time.Sleep(80 * time.Millisecond); cancel() }()
+	_ = w.Run(ctx)
+
+	if got := q.state["j1"]; got != "failed" {
+		t.Fatalf("state=%q want failed", got)
+	}
+	if q.last["j1"] == "" {
+		t.Fatalf("expected non-empty error explanation for missing image_id")
+	}
+}
+
+func TestWorker_DrainsMultipleBeforeWaiting(t *testing.T) {
+	q := newFakeQueue(
+		Job{ID: "j1", Backend: "mock"},
+		Job{ID: "j2", Backend: "mock"},
+		Job{ID: "j3", Backend: "mock"},
+	)
+	p := &fakePipeline{results: map[string]Outcome{"": {ImageID: "img"}}}
+	w := New(q, p, Config{PollInterval: 200 * time.Millisecond, JobTimeout: time.Second})
+	ctx, cancel := context.WithCancel(context.Background())
+	go func() { time.Sleep(60 * time.Millisecond); cancel() }()
+	_ = w.Run(ctx)
+
+	for _, id := range []string{"j1", "j2", "j3"} {
+		if got := q.state[id]; got != "done" {
+			t.Fatalf("%s state=%q want done", id, got)
+		}
+	}
+	if q.done != 3 {
+		t.Fatalf("done=%d want 3", q.done)
+	}
+}
+
+func TestWorker_NotifyWakesEarlierThanPoll(t *testing.T) {
+	q := newFakeQueue()
+	p := &fakePipeline{results: map[string]Outcome{"": {ImageID: "img"}}}
+	// Set poll interval high so a working LISTEN is required to see the job
+	// promptly. Without NOTIFY plumbing this test would time out the worker
+	// before drain ever runs.
+	w := New(q, p, Config{PollInterval: 5 * time.Second, JobTimeout: time.Second})
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+	done := make(chan struct{})
+	go func() {
+		_ = w.Run(ctx)
+		close(done)
+	}()
+	// Append a job and ping the wake channel.
+	q.mu.Lock()
+	q.pending = append(q.pending, Job{ID: "late", Backend: "mock"})
+	q.state["late"] = "pending"
+	q.mu.Unlock()
+	q.pingNotify()
+
+	// Give the worker a beat to claim + process.
+	deadline := time.Now().Add(500 * time.Millisecond)
+	for time.Now().Before(deadline) {
+		q.mu.Lock()
+		s := q.state["late"]
+		q.mu.Unlock()
+		if s == "done" {
+			cancel()
+			<-done
+			return
+		}
+		time.Sleep(5 * time.Millisecond)
+	}
+	t.Fatalf("worker did not pick up the late job within the 500ms window — NOTIFY wake-up path is broken")
+}
+
+func TestWorker_HonoursContextCancellation(t *testing.T) {
+	q := newFakeQueue()
+	p := &fakePipeline{results: map[string]Outcome{"": {ImageID: "img"}}}
+	w := New(q, p, Config{PollInterval: 10 * time.Millisecond, JobTimeout: time.Second})
+	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Millisecond)
+	defer cancel()
+	start := time.Now()
+	if err := w.Run(ctx); err != nil {
+		t.Fatalf("Run: %v", err)
+	}
+	if dur := time.Since(start); dur > 200*time.Millisecond {
+		t.Fatalf("worker did not exit promptly on ctx cancel: %v", dur)
+	}
+}
+
+func TestWorker_InflightJobFinishesAfterShutdown(t *testing.T) {
+	q := newFakeQueue(Job{ID: "long", Backend: "mock"})
+	p := &fakePipeline{
+		results: map[string]Outcome{"long": {ImageID: "img-long"}},
+		delay:   120 * time.Millisecond,
+	}
+	// Short JobTimeout would also kill the in-flight job; give it enough
+	// budget so the test exercises the shutdown-during-job path.
+	w := New(q, p, Config{PollInterval: 10 * time.Millisecond, JobTimeout: 5 * time.Second})
+	ctx, cancel := context.WithCancel(context.Background())
+	go func() {
+		// Let the job start, then cancel mid-flight.
+		time.Sleep(30 * time.Millisecond)
+		cancel()
+	}()
+	_ = w.Run(ctx)
+	if got := q.state["long"]; got != "done" {
+		t.Fatalf("state=%q want done (in-flight job should finish even on shutdown)", got)
+	}
+}
+
+// TestWorker_PropagatesSeriesIDToPipeline verifies the worker hands the
+// Job's SeriesID through to the pipeline unchanged. The pipeline owns the
+// cloud-sync side of the propagation (cloud.SyncRequest.SeriesID lands on
+// imagen.images.series_id) — see cloud_test.go for that half — so the
+// worker contract is simply: don't drop or rewrite SeriesID between
+// claim and Run.
+func TestWorker_PropagatesSeriesIDToPipeline(t *testing.T) {
+	const seriesID = "11111111-1111-1111-1111-111111111111"
+	q := newFakeQueue(Job{
+		ID:       "j-series",
+		Prompt:   "p",
+		Backend:  "mock",
+		SeriesID: seriesID,
+	})
+	p := &fakePipeline{results: map[string]Outcome{"j-series": {ImageID: "img-series"}}}
+	w := New(q, p, Config{PollInterval: 10 * time.Millisecond, JobTimeout: time.Second})
+	ctx, cancel := context.WithCancel(context.Background())
+	go func() { time.Sleep(80 * time.Millisecond); cancel() }()
+	if err := w.Run(ctx); err != nil {
+		t.Fatalf("Run: %v", err)
+	}
+	if got := p.lastJob.SeriesID; got != seriesID {
+		t.Fatalf("pipeline saw SeriesID=%q want %q", got, seriesID)
+	}
+	if got := q.state["j-series"]; got != "done" {
+		t.Fatalf("state=%q want done", got)
+	}
+}
+
+// TestWorker_SoloJobLeavesSeriesIDEmpty is the negative case — a job
+// claimed with no series row keeps the field empty all the way to the
+// pipeline so cloud-sync writes NULL into imagen.images.series_id.
+func TestWorker_SoloJobLeavesSeriesIDEmpty(t *testing.T) {
+	q := newFakeQueue(Job{ID: "j-solo", Prompt: "p", Backend: "mock"})
+	p := &fakePipeline{results: map[string]Outcome{"j-solo": {ImageID: "img-solo"}}}
+	w := New(q, p, Config{PollInterval: 10 * time.Millisecond, JobTimeout: time.Second})
+	ctx, cancel := context.WithCancel(context.Background())
+	go func() { time.Sleep(80 * time.Millisecond); cancel() }()
+	_ = w.Run(ctx)
+	if got := p.lastJob.SeriesID; got != "" {
+		t.Fatalf("solo job pipeline.lastJob.SeriesID=%q want empty", got)
+	}
+}
+
+func TestWorker_TransientClaimErrorDoesNotKillLoop(t *testing.T) {
+	// First claim returns an error; the loop should log and try again on the
+	// next wake — it must not propagate the error and exit.
+	q := newFakeQueue(Job{ID: "j1", Backend: "mock"})
+	q.claimErr = fmt.Errorf("transient: connection reset")
+	p := &fakePipeline{results: map[string]Outcome{"j1": {ImageID: "img"}}}
+	w := New(q, p, Config{PollInterval: 20 * time.Millisecond, JobTimeout: time.Second})
+	ctx, cancel := context.WithCancel(context.Background())
+
+	// Heal the claim error after a beat so the second drain succeeds.
+	go func() {
+		time.Sleep(40 * time.Millisecond)
+		q.mu.Lock()
+		q.claimErr = nil
+		q.mu.Unlock()
+	}()
+	go func() {
+		time.Sleep(200 * time.Millisecond)
+		cancel()
+	}()
+	if err := w.Run(ctx); err != nil {
+		t.Fatalf("Run returned: %v (transient claim errors should not kill the loop)", err)
+	}
+	if got := q.state["j1"]; got != "done" {
+		t.Fatalf("state=%q want done", got)
+	}
+}

scripts/imagen-worker.env.example

@@ -0,0 +1,22 @@
+# Environment for the imagen-worker.service systemd unit.
+# Copy to ~/.dotfiles/.env.imagen-worker and fill in real values.
+# Never commit the populated file — it carries the Supabase service-role key.
+
+# Direct Postgres DSN for LISTEN/NOTIFY + imagen.jobs UPDATE statements.
+# PostgREST cannot LISTEN, so the worker connects to Postgres directly.
+# Host + port + password come from the msupabase compose env on mlake.
+IMAGEN_WORKER_DATABASE_URL=postgres://postgres:CHANGE_ME@100.99.98.201:6789/postgres?sslmode=disable
+
+# PostgREST endpoint for the imagen.images cloud-sync writer (same as
+# `imagen generate`'s cloud-sync code path).
+SUPABASE_URL=https://supa.flexsiebels.de
+SUPABASE_SERVICE_KEY=CHANGE_ME
+
+# Default owner_user_id. Per-job owner from the imagen.jobs row overrides
+# this, so it's only used as a fallback when a job arrives with a NULL
+# owner_user_id — which the schema disallows. Keep it set for safety.
+IMAGEN_OWNER_USER_ID=ac6c9501-3757-4a6d-8b97-2cff4288382b
+
+# Optional: REPLICATE_API_TOKEN if any imagen.jobs.backend may resolve to
+# a Replicate adapter instance.
+# REPLICATE_API_TOKEN=CHANGE_ME

scripts/imagen-worker.service

@@ -0,0 +1,19 @@
+[Unit]
+Description=ImaGen worker (consumes imagen.jobs queue)
+Documentation=https://mgit.msbls.de/m/ImaGen/issues/8
+Wants=network-online.target
+After=network-online.target
+
+[Service]
+Type=simple
+ExecStart=%h/dev/ImaGen/bin/imagen worker
+WorkingDirectory=%h/dev/ImaGen
+EnvironmentFile=%h/.dotfiles/.env.imagen-worker
+Restart=on-failure
+RestartSec=5
+# Give the worker time to finish an in-flight generation on shutdown
+# (FLUX dev up to ~30s, plus the cloud-sync write-back).
+TimeoutStopSec=60
+
+[Install]
+WantedBy=default.target