agentleFS
Sign inStart free
ohad6kohad6k/emulo/.agents/skills/agent-profile-from-sessions/SKILL.md

agent-profile-from-sessions

Skill292 starsChanged 4 months agoMITView on GitHub

Use when the user keeps re-explaining the same preferences to a coding agent across chats (design taste, video style, review rules, how work should be verified, "as I said", "I told you"), or asks how to make Claude Code, Codex, Copilot CLI or OpenCode stop forgetting how they work. Runs the free local emulo --coach report over their own past sessions to show what they repeat, with dated receipts, then offers to build a reviewed you.md profile. Not for shared team memory, project documentation, or a single forgotten fact.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 6 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    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. Distinctive:

    Confirm the pattern: Ask one question: which kind of instruction keeps coming back (design, video, writing, verification, code style)?

    line 12
  2. Very specific:

    Add --source claude (or codex, copilot, opencode) to read one tool only.

    line 17
  3. Distinctive:

    Do not claim the agent will "understand" the user or follow the profile.

    line 27
  4. Distinctive:

    Do not upload, paste or summarize session logs anywhere outside the user's machine.

    line 28
  5. Distinctive:

    Do not run the mining step without showing its planned model calls and getting a yes.

    line 29
  6. Distinctive:

    Do not write the profile into a committed file without naming the file and warning that it contains quotes from the user's sessions.

    line 30
  7. Very specific:

    Do not use this for team-wide or shared memory.

    line 31
  8. Distinctive:

    Only the messages the user typed in supported local session logs (Claude Code, Codex, Copilot CLI, OpenCode, Google Antigravity).

    line 35

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

Stop repeating yourself to your coding agent

This skill turns a complaint like "I keep telling you the same thing" into evidence from the user's own history, and then into something the agent can load next time.

Phases

  1. Confirm the pattern. Ask one question: which kind of instruction keeps coming back (design, video, writing, verification, code style)? Then continue. Do not mine anything yet.
  2. Run the free report. It is local and makes no model call:
    uvx emulo@0.6.7 --coach
    Add --source claude (or codex, copilot, opencode) to read one tool only. On a large history it takes about a minute.
  3. Show the evidence. Quote the repeated asks it found, with their dates, grouped by theme. Say plainly that a text match cannot tell why something was repeated or whether the agent forgot it. The findings are possibilities, not a diagnosis.
  4. Decide the cheapest fix with the user:
    • one or two rules: suggest adding them to CLAUDE.md or AGENTS.md, or to the host's own memory, which may be enough;
    • a recurring style across many chats (for example how they want designs or films): offer a full Emulo profile.
  5. Only with an explicit yes, hand over to the emulo skill (npx skills add ohad6k/emulo@emulo). It mines the full history into a you.md after showing the planned model calls and getting cost approval.
  6. Review and install. The user reads every quote in the profile and removes anything they don't want. Then install it for the current host and verify it loads in a fresh task. In Claude Code the user types /you; in Codex, --target agents writes it into the project's AGENTS.md.

Do not

  • Do not claim the agent will "understand" the user or follow the profile. A profile improves context; it is not a guarantee.
  • Do not upload, paste or summarize session logs anywhere outside the user's machine.
  • Do not run the mining step without showing its planned model calls and getting a yes.
  • Do not write the profile into a committed file without naming the file and warning that it contains quotes from the user's sessions.
  • Do not use this for team-wide or shared memory. Emulo profiles are personal.

What it reads

Only the messages the user typed in supported local session logs (Claude Code, Codex, Copilot CLI, OpenCode, Google Antigravity). It does not read rules files, memory files or typed self-descriptions as evidence. Details: https://github.com/ohad6k/emulo and https://emulo.vercel.app/agent-profile

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