agentleFS
kuldeeps48kuldeeps48/agentic-engineering/.github/skills/update-knowledge/SKILL.md

update-knowledge

Skill0 starsChanged 7 months agoMITView on GitHub

Capture learnings from the current session and persist them across all knowledge stores: memory, skills, instructions, and command files. USE FOR: save learnings, update knowledge, persist insights, update instructions, update skills, update memory, capture session learnings, knowledge update, save what we learned, update docs from session. DO NOT USE FOR: creating new skills from scratch (use agent-customization skill), writing application code, code review (use code-review skill).

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 1

    1. jerry-20160416/skills★ 2same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository8 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    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
  1. Distinctive:

    Identify what was learned this session: Look for:

    line 17
  2. Distinctive:

    Ask the user: "What are the key learnings from this session?" if not already clear.

    line 24
  3. Distinctive:

    Both must be updated in sync: when changing always-on rules

    line 61
  4. Distinctive:

    Read the current content: before editing — check what's already there to avoid duplicates

    line 80
  5. Distinctive:

    If a new skill/command was created, update the reference tables:

    line 87

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Update Knowledge — Platform Backend

Purpose: Take learnings from the current session and update all relevant knowledge stores so the insights persist across conversations and agents.


0. Before You Start — Gather Learnings

Identify what was learned this session. Look for:

  1. Bugs found during implementation or review — patterns that were missed, edge cases discovered
  2. Rules that conflicted with behavior — structural rules that were applied mechanically and caused issues
  3. New patterns established — solutions that should be replicated in future features
  4. Corrections to existing guidance — instructions that were wrong, incomplete, or misleading

Ask the user: "What are the key learnings from this session?" if not already clear.


1. Knowledge Stores — Where to Update

There are 6 knowledge stores, each with a different purpose and format. Update only the ones where the learning is relevant.

StorePathFormatScopeWhen to Update
User memory/memories/Bullet points, briefCross-workspace, persistentGeneral workflow preferences, common mistakes
Repo memory/memories/repo/JSON (create only)Repository-scopedCodebase conventions, verified practices
CLAUDE.mdCLAUDE.mdMarkdown, uses "Slash Commands"Claude Code sessionsAlways-on rules, gotchas, patterns
copilot-instructions.md.github/copilot-instructions.mdMarkdown, uses "Skills"GitHub Copilot sessionsAlways-on rules, gotchas, patterns
Skills.github/skills/<name>/SKILL.mdMarkdown + YAML frontmatterCopilot, task-specificDeep-dive procedures, checklists
Commands.claude/commands/<name>.mdPlain markdown (NO frontmatter)Claude Code, task-specificDeep-dive procedures, checklists

Critical Format Differences

.github/skills/ (Copilot):

  • Lives in a folder: .github/skills/<skill-name>/SKILL.md
  • MUST have YAML frontmatter with name and description
  • Description should include USE FOR: and DO NOT USE FOR: keywords
  • Referenced in copilot-instructions.md Skills Reference table as [name](.github/skills/name/SKILL.md)

.claude/commands/ (Claude Code):

  • Single file: .claude/commands/<command-name>.md
  • NO YAML frontmatter — starts directly with content
  • Referenced in CLAUDE.md Slash Commands Reference table as /project:<command-name>

CLAUDE.md vs copilot-instructions.md:

  • Content is nearly identical but terminology differs:
    • CLAUDE.md says "Slash Commands" and /project:<name>
    • copilot-instructions.md says "Skills" and links to SKILL.md files
  • Both must be updated in sync when changing always-on rules

2. Update Process

Step 1: Classify the Learning

For each learning, determine:

  • Is it always-on? → Update CLAUDE.md + copilot-instructions.md (Common Gotchas, Transaction Pattern, etc.)
  • Is it task-specific? → Update the relevant skill + command (e.g., code-review, api-patterns)
  • Is it a personal workflow preference? → Update /memories/ only
  • Is it a codebase fact? → Create /memories/repo/<name>.json

Step 2: Update Each Store

For each relevant store:

  1. Read the current content before editing — check what's already there to avoid duplicates
  2. Find the right section — don't append randomly; place the learning in context
  3. Match the existing style — bullet points, numbered lists, or prose as appropriate
  4. Keep it concise — especially in memory files and always-on instructions

Step 3: Cross-Reference Tables

If a new skill/command was created, update the reference tables:

  • copilot-instructions.md → Skills Reference table
  • CLAUDE.md → Slash Commands Reference table

Step 4: Verify Consistency

  • CLAUDE.md and copilot-instructions.md should have the same rules (different format references)
  • Skills and their corresponding commands should have the same content (different frontmatter)
  • Memory should not contradict instruction files

3. Templates

New Gotcha Entry (for CLAUDE.md / copilot-instructions.md)

N. **Short title:** Explanation of the gotcha — what goes wrong and why. Include the correct approach

New Repo Memory (for /memories/repo/)

{
  "subject": "Brief title",
  "fact": "The learning in 1-2 sentences with actionable implications",
  "citations": ["file/path1.py", "file/path2.py"],
  "reason": "Why this matters for future tasks",
  "category": "architecture|convention|testing|security"
}

New User Memory Entry (for /memories/)

- **Short title:** One-line actionable insight

4. Checklist

Before finishing, verify:

  • Learnings identified and classified
  • /memories/ updated (if general workflow insight)
  • /memories/repo/ created (if codebase-specific fact)
  • CLAUDE.md updated (if always-on rule)
  • .github/copilot-instructions.md updated (if always-on rule — keep in sync with CLAUDE.md)
  • Relevant .github/skills/<name>/SKILL.md updated (if task-specific)
  • Relevant .claude/commands/<name>.md updated (if task-specific — same content, no frontmatter)
  • Reference tables updated (if new skill/command created)
  • No contradictions between stores

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.