agentleFS
LodyAILodyAI/Lody/.agents/skills/archive-agent-notes/SKILL.md

archive-agent-notes

Skill1.7k starsChanged 5 days agoApache-2.0View on GitHub

Use when adding, auditing, pruning, archiving or reviewing Lody Agent Notes. Runs the supersession check every new note requires, deletes records of small UI or purely mechanical changes, classifies implemented notes by future decision value, rejects obsolete proposals, prunes rejected notes that no longer prevent a plausible mistake, and applies the archive procedure.

Adopted by

Steady: 0 new projects took it in the last 4 weeks, none the 4 weeks before.thin data

Week of Sep 13: 0 new projectsWeek of Sep 20: 0 new projectsWeek of Sep 27: 0 new projectsWeek of Oct 4: 0 new projectsWeek of Oct 11: 0 new projectsWeek of Oct 18: 0 new projectsWeek of Oct 25: 0 new projectsWeek of Nov 1: 0 new projects
Sep 2026Oct 2026Nov 2026
New projects taking a version of this file each week, over the 8 weeks our latest run read: only 0 in all, so this is a hint, not a trend.

Who's talking about LodyAI/Lody

The biggest 5 of 50 mentions linking to the repository that our crawl found. Each line is their own headline.

See all 50

Talked about on GitHub (50 times).

On GitHub · 50

45 more on GitHub

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 7 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read the note rules and document maintenance.

    line 16
  2. Distinctive:

    Run this before drafting any note, scoped to the decision being recorded:

    line 23
  3. Distinctive:

    Search active notes for the same decision, mechanism, file paths and rejected alternatives.

    line 25
  4. Distinctive:

    If one note already owns the decision, update it.

    line 28
  5. Distinctive:

    Do not defer a known match to a later audit.

    line 37
  6. Very specific:

    Keep bug-fix/2026-09-18-codex-transient-reasoning: it shipped and was reverted, and routing transient reasoning outside history remains a …

    line 92
  7. Very specific:

    Keep architecture/2026-10-10-provider-prompt-receipts only while durable native admission receipts remain a plausible misreading of the …

    line 94
  8. Distinctive:

    Search for inbound links from active documents.

    line 102
  9. Distinctive:

    After sealing, do not edit, translate, reformat or move the pair.

    line 106
  10. Very specific:

    Run pnpm run docs check and pnpm format.

    line 110

≈ “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

Archive Lody Agent Notes

Notes are the repository's decision memory. This workflow keeps the active set (proposed/ and implemented/) small enough to search and current enough to trust, without losing history that can still guide work. Judge every note by its future decision value. Word count, age and inbound-link count are discovery aids, never criteria, and there is no quota.

Read the contracts

Read the note rules and document maintenance. Establish whether a rationale still owns or constrains anything from current code, Specs, .agents/docs/, owning READMEs, newer notes and inbound links, not from the note's own claims.

Supersession check for a new note

Run this before drafting any note, scoped to the decision being recorded:

  1. Search active notes for the same decision, mechanism, file paths and rejected alternatives. Search titles and abstracts first; git log --name-only on the changed source paths finds notes committed with earlier changes to them.
  2. If one note already owns the decision, update it. Do not create a second note.
  3. For each implemented note the new decision fully replaces, carry every unique rationale, alternative, measurement and reintroduction condition into the new note, archive the old pair, and list it under Supersedes: in the new note's metadata, linking the archived path.
  4. For a partial replacement, keep both active, cross-link them, and correct the facts in the older note that the change made stale.
  5. Reject a proposal the new decision makes obsolete; never archive a proposal.

Do not defer a known match to a later audit.

Classify by future value

Apply these outcomes per lifecycle:

  • Implemented, delete: the note records only a small UI adjustment, a local interaction fix or a purely mechanical change. Delete both languages and repair or remove inbound links. Apply this test first.
  • Implemented, keep active: its rationale, rejected alternatives, negative guarantee, durable or wire semantics, ownership boundary, security rule, measurement or reintroduction condition is likely to guide a future change. Classify the decision, not the size of its implementation.
  • Implemented, archive: a substantive decision that is complete and unlikely to guide future work, but whose rationale still has historical value.
  • Proposed, reject: the work is no longer worth pursuing or has been overtaken. Move the pair to rejected/ with the reason. A proposal whose PR merged is not obsolete; move it to implemented/ and rewrite its plan into the present tense.
  • Rejected, keep: the losing idea remains a tempting, meaningful mistake and the note explains why it loses.
  • Rejected, delete: the idea is obsolete, superseded or no longer plausible.

Classify analogous notes under one principle, use judgement for close cases, and report genuinely borderline ones. A note kept active must end the change linked from its current-state owner (README, .agents/docs/ page, AGENTS.md or Spec) with one sentence of what is true now; pnpm run docs status lists active notes that lack one (unlinkedNotes) and is the discovery list for a topic audit.

Calibrated examples

These set the bar. The sizes show that length is not the test.

Delete:

  • bug-fix/2026-09-21-model-search-focus (388 words): a local menu focus fix; its regression lives in the browser suite and the run-config doc owns the overview.

Keep active:

  • bug-fix/2026-10-22-node-security-runtime-floor (598 words): the root AGENTS.md cites it as the rationale for a binding version floor.
  • architecture/2026-10-12-operation-inbox (3931 words): an ownership and admission contract other notes and AGENTS.md files depend on.

Archive:

  • bug-fix/2026-10-27-e2ee-cli-review-fixes (512 words), once the E2EE CLI decisions it closes are owned by a current note or .agents/docs/ page: it is a review-fix batch whose lasting rules belong to those owners.
  • simplification/2026-09-10-goal-control-ablation (20871 words): the adopted simplifications shipped; its per-file ablation matrix is historical evidence, not current guidance.

Rejected:

  • Keep bug-fix/2026-09-18-codex-transient-reasoning: it shipped and was reverted, and routing transient reasoning outside history remains a tempting shortcut.
  • Keep architecture/2026-10-10-provider-prompt-receipts only while durable native admission receipts remain a plausible misreading of the delivery request.

Archive one implemented pair

  1. Move foo.md and foo.zh.md from implemented/<type>/ to archived/<type>/.
  2. Make no body edits. Insert the same Archived: YYYY-MM-DD line immediately below Status: implemented in both files.
  3. Search for inbound links from active documents. Redirect them to the current owner, retarget them to the archived path only where history is cited on purpose, or delete them. Link repair inside the archived note itself is allowed only for a broken target, never to update its facts.
  4. After sealing, do not edit, translate, reformat or move the pair.

Validate and report

Run pnpm run docs check and pnpm format. Report notes kept, deleted, archived and rejected, and every borderline case with the outcome you chose.

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.