القائمة
How to review proposed skills, curate shared memory, and detect patterns.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Complete reference for the factory CLI — all commands, arguments, options, and usage. Use when an agent needs to invoke factory commands, check correct syntax, or understand available options.
When and how to write a human-action-needed entry to memory/{agent}/needs.md. Agents write these entries when they hit blockers that only a human can resolve.
When you encounter a failure, correction, or new pattern, write a learning and optionally propose a skill.
How the factory workspace is organized and what each directory means.
What to examine during a reflection pass and how to write observations to memory/{agent}/needs.md.
When and how to write scenario holdouts for factory verification.
| name | knowledge-curation |
| description | How to review proposed skills, curate shared memory, and detect patterns. |
On every heartbeat, check skills/proposed/ for pending skill files. This is your primary intake queue — agents propose skills here and depend on you to review and promote them. If skills/proposed/ has entries, reviewing them is your first priority.
universe/?skills/shared/?If it passes: move to skills/shared/, commit with a note about what was reviewed.
If it needs work: edit it in skills/proposed/, leave a note about what changed.
If it's redundant or wrong: delete it, write a brief explanation in learnings/.
Scan learnings/failures/ for recurring themes. If Builder hits the same class of problem 3+ times, that's a pattern. Write a skill for it.
Scan daily logs across agents for repeated workarounds. If agents keep doing the same manual step, that's a skill waiting to be written.
tasks/verified/ and specs/archive/, populating it is the priority. Survey verified tasks and archived specs to extract conventions, patterns, and operational facts that belong in shared knowledge. An empty KNOWLEDGE.md with 12 verified tasks means curation hasn't happened yet — not that there's nothing to curate.When reviewing skills, cross-reference with universe/context-engineering/ to ensure proposed skills align with context discipline principles. A proposed skill that encourages loading too much context is harmful regardless of its content.