agentleFS
Sign inStart free
Soe-RehSoe-Reh/soes-skills/skills/learn/SKILL.md

learn

Skill1 starsChanged 4 days agoMITView on GitHub

Use when a session or project wraps up, or the user says "learn", "what did we learn", "retro", "self-improve", "remember this for next time", or complains that the agent looped, retried, or burned too many tokens or too much time.

194 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use add-anydoc, add-clidash and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • TomGranotSkill

    add-anydoc

    Add local office-document-to-Markdown conversion to NanoClaw agent containers with the pinned Firecrawl AnyDoc CLI. Use when agents need to read attached Word, PowerPoint, Excel, OpenDocument, RTF, EPUB, CSV, or text-based PDF files without uploading them to a hosted parser.

    Also in 5 projects that use this

  • TomGranotSkill

    add-clidash

    Add clidash — a zero-dependency, read-only web dashboard that derives its tabs and tables at runtime from any CLI that lists resources as JSON. Ships pre-wired for NanoClaw's ncl CLI (agent groups, sessions, channels, users, roles), plus message-activity charts, a log tail, and a read-only file viewer for group skills/CLAUDE.md/profiles.

    Also in 5 projects that use this

  • seek888Skill

    open-gstack-browser

    Launch GStack Browser — AI-controlled Chromium with the sidebar extension baked in. Opens a visible browser window where you can watch every action in real time. The sidebar shows a live activity feed and chat. Anti-bot stealth built in. Use when asked to "open gstack browser", "launch browser", "connect chrome", "open chrome", "real browser", "launch chrome", "side panel", or "control my browser". Voice triggers (speech-to-text aliases): "show me the browser".

    Also in 11 projects that use this

  • Saqlain-811Skill

    ios-sync

    Regenerate the iOS debug bridge against the latest upstream gstack templates. (gstack)

    Also in 6 projects that use this

  • MicrosoftSkill

    add-codex

    Use Codex (CLI + AppServer) as the full agent provider — planning, tool orchestration, native compaction, MCP tools, session resume — in place of the Claude Agent SDK. ChatGPT subscription or OPENAI_API_KEY. Per-group via agent_provider. Distinct from using OpenAI as an MCP tool (where Claude remains the planner).

    Also in 6 projects that use this

  • MicrosoftSkill

    add-deltachat

    Add DeltaChat channel integration via @deltachat/stdio-rpc-server. Native adapter — no Chat SDK bridge. Email-based messaging with end-to-end encryption.

    Also in 6 projects that use this

  • +5 more from TomGranot
  • +1 more from microsoft

The file

We split this skill into the rules it gives an agent

2 rules. 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 169 edited versions we read keep these.

  • 0

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use the current conversation.

    line 12
  2. Very specific:

    Read the homes once: project CLAUDE.md (and nested ones), the memory index MEMORY.md, and the description line of each skill in this plugin.

    line 13

“Common” is measured: kept by at least 6 in 10 of the 169 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

Learn

Turn this session into a few durable rules, each in the one place that will apply it next time. Keep this pass cheap.

1. Gather evidence

  1. Use the current conversation. If it starts from a summary, read only the compacted part of the session transcript (newest .jsonl under ~/.claude/projects/<project>/) by grepping user messages.
  2. Read the homes once: project CLAUDE.md (and nested ones), the memory index MEMORY.md, and the description line of each skill in this plugin. Open a skill body only when a lesson points at it.

2. Scan, highest value first

  1. Corrections: the user said no, stop, actually, or fixed your work.
  2. Waste: the same command, fix, file read, or subagent run three or more times; a review loop that ran rounds past the bar; a big read that a grep would have answered. Count it and fix the cause (missing info, vague stop rule, no check).
  3. Repeated asks: guidance the user gave twice.
  4. Confirmed approaches: something that worked and will be needed again.
  5. Failures: what broke, and what worked instead.
  6. Preferences: style, naming, tool choices the user stated.

3. Filter

Keep a lesson only if it is stable, non-obvious, actionable as one rule, and not already covered. It is covered when an existing rule, followed, would have prevented the problem. A rule that exists but was ignored needs to be sharper or moved, not copied. Drop one-offs, task state, and bugs fixed this session. If nothing survives, say so and stop.

4. Route each lesson to one home

HomeGets
Automated check (lint, hook, test)Mechanical rules a tool can decide from the code. Write no prose rule for these.
A skill in this pluginAnything that corrects or sharpens how a skill runs: a skipped step, a bad default, a missing stop rule.
Project CLAUDE.mdProject decisions and pointers (commands, paths, conventions). Keep it short; it loads every session.
User memoryPersonal preferences and gotchas with no skill home.

Ties: skill beats CLAUDE.md and memory. Decisions go to CLAUDE.md, discovered quirks to memory. A multi-step procedure goes to a skill, with a one-line pointer in CLAUDE.md.

5. Show the plan, then write

  1. Print a table: lesson, exact file, append or edit in place, and the one-line rule text. Wait for approval. Write nothing before it.
  2. Apply only approved rows. Edit in place over appending, match the file's tone, state the rule, not the incident. Keep personal details out of files that are shared or published.
  3. Report what changed in one line per file.

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