agentleFS
Rome-1Rome-1/obsidian-claude-starter/.claude/skills/log-feedback/SKILL.md

log-feedback

Skill5 starsChanged 33 days agoMITView on GitHub

Captures friction with the vault's own system (a wrong or stale instruction in AGENTS.md, a template missing a field, a skill that fired when it should not have or gave bad guidance, a correction the user made to an agent's work) and fixes the cause, not just the instance; appends to System/agent-feedback.md and, when the fix is clear, edits the skill, template or rule in the same pass. Use when the user says "remember this for next time", "don't do that again", "that instruction is wrong", "update the skill", "log this", corrects the same kind of mistake twice, runs /log-feedback, or when you notice an instruction here contradicting what actually works.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 4 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    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:

    When the user corrects an agent, two things need fixing: the output in front of you, and whatever produced it.

    line 9
  2. Distinctive:

    If you cannot name a source or a gap, ask one question to find it.

    line 20
  3. Distinctive:

    Ask: what rule, if it had been written down, would have prevented this, and does that rule hold beyond this one case?

    line 28
  4. Distinctive:

    Prefer a rule that generalizes ("cut any qualifier the sentence survives without") over a patch for one word ("don't say 'very'").

    line 29
  5. Very specific:

    When resolving an older open entry, append a new line (- YYYY-MM-DD [re: YYYY-MM-DD entry] applied | declined: ...) instead of editing it.

    line 52
  6. Very specific:

    Commit the fix and the log line together, e.g.

    line 56

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

Log feedback

When the user corrects an agent, two things need fixing: the output in front of you, and whatever produced it. Fixing only the first means the same mistake comes back next session.

1. Pin it down

Get a concrete pointer, not an impression:

  • What happened (the output, or the instruction that misled).
  • Where it came from: a line in AGENTS.md, a section of a skill, a template field, the writing guide, or nowhere (a gap).
  • What should have happened instead.

If you cannot name a source or a gap, ask one question to find it.

2. Fix the instance

Correct the note, reply or file the user is unhappy with, if that has not been done.

3. Find the general lesson

Ask: what rule, if it had been written down, would have prevented this, and does that rule hold beyond this one case? Prefer a rule that generalizes ("cut any qualifier the sentence survives without") over a patch for one word ("don't say 'very'").

4. Route it

The lesson is about...Put it in
How a specific workflow should runThat skill's SKILL.md (a step, a check, or a sharper trigger in the description)
A field or section a note type needsThe template in System/Templates/ and [[System/vault-schema]]
How prose should readSystem/writing-guide.md, under "Your additions", with an example
How agents should behave in this vaultAGENTS.md. Keep it short; it loads every session
How agents should behave everywhereThe user's global rules (see System/global-agent-rules.md); propose the edit, since that file lives outside the vault
Unclear, or needs the user to decideOnly the log entry below

Make small, additive edits. Do not rewrite a skill or a rule file wholesale to fix one thing. For edits to AGENTS.md or global rules, show the user the diff before saving.

5. Log it

Append one line to System/agent-feedback.md under ## Entries:

  • Fixed now: - YYYY-MM-DD [where] what went wrong; fixed in <file> [status: applied]
  • Not fixed: - YYYY-MM-DD [where] what went wrong, and the proposed fix [status: open]

When resolving an older open entry, append a new line (- YYYY-MM-DD [re: YYYY-MM-DD entry] applied | declined: ...) instead of editing it.

6. Commit

Commit the fix and the log line together, e.g. vault: feedback - meeting-notes asks for owners instead of guessing, if the user has agreed to agent commits.

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.