agentleFS
Sign inStart free
dakldakl/engram/.claude/skills/self-reflection/SKILL.md

self-reflection

Skill2 starsChanged 5 months agoMITView on GitHub

Reflect on `CLAUDE.md` and the skill files under `.claude/skills/` after a working session — surface rules that helped, rules that produced friction, rules that were skipped, and stale references. Edit the ruleset directly; commit as `docs(claude): self-reflection ...`. Activates when the user types `/self-reflection` or `/reflect`, when `/work` terminates, or whenever a rule feels wrong mid-session.

6 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions6 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 6

    1. QuixiAI/agi-memory★ 623same name
    2. QuixiAI/Hexis★ 623same name
    3. dglijin-oss/chinese-metaphysics-skills★ 13same name
    4. rolandbrecht/agent-skills★ 4same name
    5. jiesou/skills★ 1same name
    6. Bot2353/skills★ 0same 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 repository5 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

8 rules. 0 are common, 8 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 5 edited versions we read keep these.

  • 8

    Distinctive

    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:

    After a session where a rule felt wrong — capture immediately.

    line 15
  2. Distinctive:

    After a change that may have invalidated rule references (a renamed module, a new ADR, a changed CLI surface).

    line 16
  3. Distinctive:

    If the session had no friction and no stale references, write nothing.

    line 19
  4. Distinctive:

    Did it actively help?: Keep it.

    line 31
  5. Distinctive:

    Did a new pattern appear?: Write the rule now. Be specific — name the file, the failure mode, the recurring brief instruction.

    line 35
  6. Distinctive:

    Link, don't duplicate: If a rule lives in another file, link to it.

    line 42
  7. Distinctive:

    Don't fabricate observations: If a rule wasn't tested this session, don't touch it.

    line 43
  8. Distinctive:

    If no edits are warranted, skip the commit and say "no changes warranted."

    line 54

“Common” is measured: kept by at least 6 in 10 of the 5 edited versions of this file we could read. “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

/self-reflection — keep the ruleset current

CLAUDE.md and the skill files under .claude/skills/ are living documents. Stale rules and silently-skipped rules degrade them. This skill walks the ruleset and updates it based on what was just observed.

0. When to run

  • At the end of a /work run (the work skill invokes this).
  • After a session where a rule felt wrong — capture immediately.
  • After a change that may have invalidated rule references (a renamed module, a new ADR, a changed CLI surface).
  • On user demand via /self-reflection or /reflect.

If the session had no friction and no stale references, write nothing. A null reflection is valid.

1. Read the surface

  • CLAUDE.md (project) — the authoritative ruleset.
  • .claude/skills/work/SKILL.md, deslop/SKILL.md, queue/SKILL.md — skill rules.
  • git log --oneline -20 — what just happened.

Narrow to the user's hint if provided.

2. Ask of each rule

  • Did it actively help? Keep it.
  • Did it get in the way? Fix the rule or document the exception + when to apply it.
  • Was it skipped? Why? Add a clearer threshold/trigger, or delete it.
  • Is anything stale? File paths, module names, CLI subcommands, ADR numbers, commit IDs, references that no longer exist. Update or delete.
  • Did a new pattern appear? Write the rule now. Be specific — name the file, the failure mode, the recurring brief instruction.

3. Edit

Edit CLAUDE.md and skill files directly. Same-session edits — future sessions won't have your in-memory exception list.

  • Tighten, don't expand. One sentence beats a paragraph.
  • Link, don't duplicate. If a rule lives in another file, link to it.
  • Don't fabricate observations. If a rule wasn't tested this session, don't touch it.
  • Architectural learnings go in an ADR, not just CLAUDE.md, when they change a decision.

4. Commit

docs(claude): self-reflection after <what just happened>

Body: substantive changes, one line each. Inspect git diff --cached --stat — reflection commits should only touch CLAUDE.md and .claude/skills/** (and an ADR if a decision changed).

If no edits are warranted, skip the commit and say "no changes warranted."

What you must NOT do

  • Fabricate edits to look productive. Null reflection is valid.
  • Rewrite a rule just because you didn't like the phrasing.
  • Auto-fix mid-task. Save reflection for a natural pause.
  • Reflect on rules that aren't yet tested.
  • Bundle unrelated edits. Reflection commits touch only the ruleset (+ ADR if warranted).
  • Delete a rule because it was inconvenient once. One data point isn't enough.

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.

Check your setup, free