agentleFS
Sign inStart free
mcabreradevmcabreradev/hermes-sdd-agency/skills/project-learnings/SKILL.md

project-learnings

Skill1 starsChanged 57 days agoMITView on GitHub

Persistent per-project learnings, versioned with the code, not global.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is 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.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep a durable, per-project record of what the project has learned, versioned in the repo so it travels with the code and the PR — …

    line 9
  2. Very specific:

    Review / search: read the file, filter by kind / area, or search text.

    line 37
  3. Distinctive:

    Keep entries atomic and one-line-summarized.

    line 46

≈ “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
<!-- hermes-add-skill: pattern adapted from gstack /learn (garrytan/gstack, MIT). -->

Project learnings

Keep a durable, per-project record of what the project has learned, versioned in the repo so it travels with the code and the PR — complementing (never duplicating) the agent's global memory and the vault. What is specific to this project belongs here; what is global stays in global memory.

Storage

.context/learnings.jsonl in the project root — one JSON object per line.

Each entry:

{"ts": "…", "kind": "lesson|pitfall|decision|pattern", "area": "<module/subsystem>", "summary": "<one line>", "detail": "<why — the 'because' that survives the session>"}

kind keeps them filterable; area ties them to code. The detail is the non-obvious why (a gotcha, a workaround, an invariant) — the thing a future session would re-learn the hard way.

When to write

  • A solution to a bug that took >1 attempt (record the root cause, not the symptom).
  • A non-obvious project invariant or constraint that code doesn't show.
  • A tooling / environment gotcha specific to this project.
  • A decision that would otherwise be re-litigated.

Managing

  • Review / search: read the file, filter by kind / area, or search text.
  • Prune: drop entries that are stale or superseded; keep the signal.
  • Export: dump the current set where the user needs it.

Rules

  • Never commit a secret, a path with credentials, or live data into a learning.
  • Learnings are committed as part of the change they belong to (they go in the PR), so they are peer-reviewed with the code.
  • Keep entries atomic and one-line-summarized; the file is a signal ledger, not a log.

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