# Dircreative Learn

> Record reusable project learnings, prompt fixes, channel rules, and model behavior.

- Skill: `papperrollinggery/dircreative-learn` (Agent Skill)
- Install (CLI): `npx skillmds@latest add papperrollinggery/dircreative-learn`
- Raw SKILL.md: https://api.skillmd.com/api/skills/papperrollinggery/dircreative-learn/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: papperrollinggery (https://skillmd.com/u/papperrollinggery)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/papperrollinggery/dircreative-learn

---


# Learn

## Required Knowledge

Read only the reference needed for the active task, not this entire list.
The root v2 route owns scope and authorization; legacy records do not add gates.

- `docs/film-preproduction/prompt-pattern-registry.json`
- `.dircreative/learnings.jsonl`
- `.dircreative/timeline.jsonl`

## Inputs

- QA report
- retry plan
- user feedback
- observed output behavior
- evidence-bound reference analysis and effect-comparison review

## Outputs

- learning entry
- registry update request
- deprecation request when a pattern fails repeatedly

## Rules

- Write append-only JSONL.
- Mark source as `user-stated`, `observed`, `inferred`, `official`, or `internal_eval`.
- Do not promote a pattern without fixture QA improvement.
- Reference mechanisms remain candidates until an original transfer test shows
  the intended audience effect without copying distinctive expression. Use
  `skills/dircreative/references/video-distillation.md` for scoped research, before/after evidence,
  review, and source-maintenance boundaries; no autonomous global installation.

## skill_run_receipt

Persist the following only for a requested formal handoff, pause/resume or actual
execution record. Ordinary work returns its result without a separate receipt.
The next skill is advisory; the controller continues only the requested scope.

Record new learning keys, confidence, source, and `next_recommended_skill: checkpoint`.