papperrollinggery/Paperrolling-DIRcreative-SKILL / skills/dircreative
dircreative
VerifiedUse after explicit $dircreative for film concepts, stories, scripts, director or creative groups, shot/asset planning, reference-video analysis, and image/video prompts. Real media execution requires the corresponding authorization. Validated ADCO handoffs are supported. DIRcreative source maintenance is outside this Skill.
references/image-generation-diagnostics.md
View on GitHub# Image generation diagnostics without expanding the workflow
Keep the user's authored body, clothing and style facts. One failed output is
not evidence that a neckline, adult body type or garment category is universally
unsupported. Do not add a prohibition list or silently close the garment.
- Distinguish `moderation_blocked` from quota, transport and server failures.
Record the returned request ID and optional `moderation_stage`/categories.
`output` describes a generated-result check, not a diagnosis of a forbidden
word. No returned image means no visual-quality verdict on that result.
- Prefer the user's complete tested prompt as a baseline. Do not automatically
translate, expand it, or layer a second style capsule over an already explicit
costume. Necessary reference roles and actual generation arguments remain
recorded; remove duplicate protocol prose, not the requested design.
- A useful bounded comparison keeps the tool and references fixed and compares
authored content with the actual compiled content. Preserve both outcomes.
One success and one block establish a difference in observed results, not a
causal explanation, a general success rate or a guaranteed remedy.
- Do not automatically retry moderation user errors, change accounts or lower
safeguards. If a non-explicit request has an unexpected result, inspect the
actual request for ambiguity/conflicts and use a genuinely appropriate revised
request or a documented support report. Do not silently change the requested
output and call that a successful test of the original.
- The native tool's supported arguments are authoritative. A parameter in an
API guide does not make it available in this host, nor prove a different
product entrance uses the same generation/monitor configuration.
Sources checked 2026-09-08:
- [Official image API guide](https://developers.openai.com/api/docs/guides/image-generation):
distinguishes input/output moderation and user-correctable errors.
- [Images 2.0 system card, 2026-04-21](https://deploymentsafety.openai.com/chatgpt-images-2-0/automated-evaluations-and-adversarial-testing):
describes upstream and downstream checks. It does not explain a particular
request's failure.
- [Community first-person reproduction, 2026-08-27](https://community.openai.com/t/inconsistent-image-safety-blocks-ordinary-adult-swimwear-repeatedly-classified-as-sexual/1392919):
reports inconsistent ordinary adult fashion creation/edit outcomes.
- [Community observation, 2026-09-07](https://community.openai.com/t/why-is-create-an-image-of-apple-sometimes-blocked-for-nudity-sexuality-has-anyone-else-seen-this/1395517):
proposes context or temporary-state effects. This is not a confirmed service
root cause. Do not use community speculation as an instruction or capability fact.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |