Agents fail generation QA in two symmetric ways: accepting the first roll, or rewording the whole prompt and re-rolling until the budget dies. Both skip the same two artifacts, a written rubric and a diagnosis. The loop that converges: rubric before generating, a small batch, a recorded verdict per asset, the cheapest targeted fix per failure, a hard round cap. Connection and the core loop: see the scenario skill. Critic tool contracts: scenario-asset-analysis. Baseline discipline: scenario-consistency. If a sibling skill named here is missing from your available skills, ask the user to install it (npx skills add scenario-labs/skills --skill <name>); unattended, proceed from tool schemas and flag the gap.
| Step | \nDo | \n
|---|---|
| 1. Rubric | \nBefore generating, turn the brief into pass/fail lines a viewer can check (\"subject centered on a plain field\"), never taste words | \n
| 2. Generate | \nThe smallest batch that tests the recipe, at the cheapest size or quality tier the schema offers on which every rubric line can still be judged; dry_run when cost matters | \n
| 3. Critique | \nasset_analyze: up to 10 images per call, one instruction embedding the rubric and a fixed per-image output shape | \n
| 4. Fix | \nRoute every fail line to the cheapest fix that addresses it (table below) | \n
| 5. Stop | \nA clean round ships; three rounds without one, or one line failing twice under different fixes, means report, not respin | \n
A tier change is a new generation, not the keeper enlarged: re-run only the keeper's recipe at delivery tier and re-critique it, or keep what passed and upscale it with a fidelity upscaler found by search target=\"models\", filters={\"tags\": [\"image-upscale\"]}, public=true (fidelity versus creative picks and sizing in scenario-image-editing, video upscaling in scenario-video).
When the bar is the configured brand brief rather than a task rubric, and the team's Quality Gate add-on is enabled, critique images with asset_quality_gate_run instead: its reasons and suggestions feed the fix table directly (scenario-quality-gate; where the gate is missing it degrades to this asset_analyze path).
Fix routing, cheapest first:
\n| The verdict says | \nFix | \n
|---|---|
| One local defect on a keeper | \nMasked inpaint of that region; on a schema with no mask field, an instruction edit of the keeper naming that one change (scenario-image) | \n
| A uniform finish off (grade, tint, crop) | \nA deterministic tool pass (scenario-image-editing), not a re-roll | \n
| Wrong content, composition, or rendered palette | \nEdit the delta clause, re-run from the approved baseline | \n
| Identity or style drift | \nTighten the enumeration, add or re-role references (scenario-consistency) | \n
| Every line failing | \nChange the model: re-discover with recommend (capability-shaped; search is for a name, a private model, or a tag-filtered lane such as image-upscale), keep the prompt | \n
Change one variable per round. A round that swaps prompt, references, and model at once cannot attribute the improvement, so the next failure restarts from zero.
\n<index>: pass|fail, <the failed line>. \"Could be better\" is not a verdict; a loop chasing better instead of the brief sands off exactly what made the direction distinctive and converges on generic output.scenario-consistency explains why).model_run each per scenario-image, then jobs_wait.asset_analyze call with all four ids in images and the rubric-plus-shape instruction; answers land as text assets, asset_download them to read the verdicts.jobs_wait the fix runs, then re-critique only the two new assets with the byte-identical instruction. Clean round: stop, file the keepers in a collection (scenario-asset-analysis).asset_display in chat: unrecorded impressions do not accumulate; verdicts do.asset_analyze to improve or fix the image: it returns text only; every fix is a new run.dry_run prices the next round ahead); usage totals lag and answer the report after the run, not the mid-run gate.