agentleFS
gabrielbchacongabrielbchacon/ai-dev-bootstrap/skills/kb-note/SKILL.md

kb-note

Skill0 starsChanged 3 months agoMITView on GitHub

Writes a knowledge base note - a decision (ADR), an incident or a research note - using the template already in the target folder, with the required frontmatter and naming, and then offers a surgical update to kb/index.md. Do NOT use to found the knowledge base itself (that is project-bootstrap). Examples of when to reach for it: - "record this decision" -> kb-note - "log this bug properly" -> kb-note - "write up what we found about the two libraries" -> kb-note

2 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions2 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 2

    1. walkwithedi/evergreen-wiki★ 2same name
    2. walkwithedi/drive-wiki★ 1same 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 repository4 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 5 are distinctive, and 4 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.

  • 4

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If the answer is "none of these", it probably does not belong in the knowledge base.

    line 25
  2. Very specific:

    Read kb/<folder>/_template.md before writing.

    line 30
  3. Distinctive:

    Do not reconstruct the structure from memory, even if you have written one before.

    line 34
  4. Distinctive:

    Ask for what is missing rather than filling it in plausibly:

    line 38
  5. Very specific:

    Leave it open instead.

    line 50
  6. Very specific:

    Write in the language the project uses for its notes — check an existing note or bootstrap.yaml.

    line 61
  7. Distinctive:

    Only after the note exists, and only as a surgical edit to the one section that changed.

    line 66
  8. Distinctive:

    Never write a whole file without reading the existing one first.

    line 76
  9. Very specific:

    Before setting status: resolved on anything involving concurrency, timing or reproducibility: an independent re-run is required.

    line 83

≈ “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

Writing a knowledge base note

1. Pick the type

TypeFolderUse when
Decision (ADR)kb/decisions/a choice was made between real alternatives
Incidentkb/incidents/something broke, or behaved unexpectedly
Researchkb/research/options were compared, or a question investigated

If the answer is "none of these", it probably does not belong in the knowledge base. Something derivable from the code or from git history is not a note.

2. Read the template first

Read kb/<folder>/_template.md before writing. Every folder has one, and the sections in it are not suggestions — the queries in index.md and the checks in verify.py both depend on the shape.

Do not reconstruct the structure from memory, even if you have written one before.

3. Gather what the template needs

Ask for what is missing rather than filling it in plausibly:

  • Decision — context, the decision itself, the alternatives with the specific reason each lost, consequences. An alternative listed without a reason is decoration.
  • Incident — symptom exactly as observed, root cause (or open if genuinely unknown), the fix, and something executable that prevents recurrence. "Be more careful" is not prevention.
  • Research — the question, findings each attributed to a source, then the recommendation in its own labelled section.

A guess recorded as a root cause is how the same incident gets investigated twice. Leave it open instead.

4. Name and write it

kb/<folder>/<slug>.md, date-prefixed where the type is dated by nature. Research notes are named by domain instead. Same-day collisions get a numeric suffix — never overwrite.

Frontmatter is mandatory: date, tags, status, plus related_spec when a spec applies (omit the key entirely when it does not).

Write in the language the project uses for its notes — check an existing note or bootstrap.yaml.

5. Then offer the index update

Only after the note exists, and only as a surgical edit to the one section that changed. Show the user the exact line you propose adding and let them approve it.

kb/index.md gets a snapshot of two or three lines and a link — never a copy of the note's reasoning. If the project has Dataview enabled, the note may already appear in the generated lists, in which case no manual edit is needed at all; check before proposing one.

Rules

  • Never write a whole file without reading the existing one first.
  • A published note is appended to, never rewritten: **Update YYYY-MM-DD**: <what changed>.
  • A superseding decision is a new note that says what it supersedes; the old one gains a forward pointer and keeps its content.
  • Resolved items in living lists are struck through, not deleted.
  • Link with [[wikilinks]] instead of repeating text across notes.
  • Before setting status: resolved on anything involving concurrency, timing or reproducibility: an independent re-run is required. A self-report is not evidence.

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.