knowledge-extract
Skill1 starsChanged 23 days agoMITView on GitHub
Turn a repeatedly observed behavior into a skill draft with a trigger, procedure, anti-patterns, and checks. Use when the same correction or workflow has appeared at least three times.
In this repository46 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .codex/AGENTS.mdAGENTS.md
- .gemini/AGENTS.mdAGENTS.md
- .cursor/rules/agent-foundry-orchestrator.mdcCursor rule
- .cursor/rules/agent-foundry-skills.mdcCursor rule
- web/llms.txtllms.txt
- anti-slopSkill
- api-designSkill
- automation-pickSkill
- bottleneck-gatingSkill
- budget-guardSkill
- cache-identitySkill
- constraint-then-solveSkill
- context-optimizationSkill
- cron-troubleshootSkill
- e2e-test-strategySkill
- engram-routingSkill
- execution-observabilitySkill
- fallback-model-routingSkill
- feedback-loopSkill
- judge-integritySkill
- landscape-firstSkill
- measure-firstSkill
- plan-before-codeSkill
- plan-then-actSkill
- prompt-disciplineSkill
- pushback-when-wrongSkill
- quality-protocolSkill
- read-before-buildSkill
- re-verify-findingsSkill
- session-closeoutSkill
- session-distillSkill
- show-your-workSkill
- structured-output-contractSkill
- verify-firstSkill
- workflow-decomposeSkill
- agent-memory-vaultSkill
- agent-team-standardsSkill
- autonomous-briefingSkill
- book-to-skillSkill
- chrome-devtools-mcp-bridgeSkill
- design-languageSkill
- funnel-pr-guardSkill
- persistent-memorySkill
- sql-migration-trioSkill
- token-compressionSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 57
If the note is still interesting a week later, draft the skill.
- Distinctive: line 81
Do not invent a script, hook, or companion dependency.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Knowledge Extract
Turn observed behavior into a reusable skill draft. Drafts stay under the
private plans/ tree until a human promotes them.
Procedure
Pipeline
Observe -> Capture -> Draft -> Validate -> Promote| Step | Output | Location |
| Observe | Pattern seen at least three times | session notes |
| Capture | One-paragraph note | plans/skills/drafts/notes/ |
| Draft | Full SKILL.md | plans/skills/drafts/<name>/SKILL.md |
| Validate | Evaluator and validator pass | same draft folder |
| Promote | Human moves the folder | skills/core/<name>/ |
Step 1 - Observe
A pattern qualifies when it repeats:
- The same correction was applied to two different problems.
- The same pre-check precedes a class of work every time.
- A recurring phrasing maps to a recurring intent.
One occurrence is an anecdote. Wait for the third.
Step 2 - Capture
Draft one paragraph in a note:
## Pattern
[One sentence: what recurs]
## Why it matters
[The cost when the pattern is missed]
## Examples
- [date] [context where it applied]If the note is still interesting a week later, draft the skill.
Step 3 - Draft
Create plans/skills/drafts/<name>/SKILL.md with:
---
name: <kebab-name>
description: "<what it does>. Use when <specific situation>."
version: 0.1.0
license: MIT
author: Agent Foundry Contributors
status: draft
source: local
---Then add:
- A procedure with concrete steps.
- At least two anti-patterns with reasons.
- A checklist that tests this skill's output.
Do not invent a script, hook, or companion dependency. If the pattern needs one, say it does not ship and describe the manual fallback.
Step 4 - Validate
./.venv/Scripts/python.exe scripts/foundry-eval.py --target plans/skills/drafts/<name>/SKILL.md
bash scripts/validate.shBoth must pass. The evaluator checks metadata, trigger count, path references, and body size.
Step 5 - Promote
A human reviews the draft, changes status: draft to status: active, moves
the folder to skills/core/<name>/, and regenerates the skill index. No script
promotes drafts automatically.
Anti-patterns
- Drafting from one occurrence.
- Promoting a draft into
skills/while its claims are still unverified. - Writing a generic checklist that could belong to any skill.
- Referencing a helper that does not exist.
- Keeping the draft path outside the private
plans/tree.
References
- details
Verification Checklist
- The pattern has at least three observed occurrences.
- The draft lives under
plans/skills/drafts/. - The description has exactly one trigger phrase.
- The checklist tests this skill's actual output.
foundry-evalpasses before promotion.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

