agentleFS
xialangji-debugxialangji-debug/codex-skills-agentmd-obsidian/skills/obsidian-fix-pattern-memory/SKILL.md

obsidian-fix-pattern-memory

Skill3 starsChanged 2 months agoMITView on GitHub

Search or maintain canonical Obsidian fix-patterns and exact-target evidence, or report validation debt and Campaigns. Use for 读取记忆库, 根据记忆, 记一下, behavior-changing fixes, regression lookup, evidence/status updates, 待真机, 验证债务, 待回归, Campaign, or 发布前验收清单. Routine recording uses one direct path; reporting and maintenance are explicit modes.

The file

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 5 are distinctive, and 6 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.

  • 6

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use scripts/fix_memory.py as the only canonical-note writer.

    line 14
  2. Distinctive:

    After a behavior-changing fix, consume the current Bug receipt:

    line 19
  3. Very specific:

    If the receipt already records this fix and verification in an exact canonical target, reuse it without another write.

    line 21
  4. Very specific:

    If memory_target is empty, search fix-patterns/ once with one to three root-cause, symbol, or symptom terms.

    line 24
  5. Distinctive:

    Update one high-confidence root-cause match.

    line 26
  6. Distinctive:

    Keep ambiguous roots separate instead of merging by project/customer name.

    line 27
  7. Very specific:

    Stop after the write. Do not run full validate, migration, domain audit, or cross-project candidates during routine recording.

    line 28
  8. Distinctive:

    Update only the exact target row:

    line 52
  9. Very specific:

    Use project/branch/version/domain filters and --campaign for the requested scope.

    line 71
  10. Very specific:

    After a write, report the note path, fix_id, exact target state, evidence level, and whether any other targets remain candidates.

    line 85
  11. Distinctive:

    Read fix record schema only when interpreting identity/state transitions.

    line 88

≈ “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

Obsidian Fix-Pattern Memory

Canonical notes live under:

%USERPROFILE%\Documents\Obsidian\CodexVault\Codex\fix-patterns

Use scripts/fix_memory.py as the only canonical-note writer. Never use Basic Memory MCP or invent a second state schema.

Fast Record Path

After a behavior-changing fix, consume the current Bug receipt:

  1. If the receipt already records this fix and verification in an exact canonical target, reuse it without another write. For new evidence, call upsert --note or the matching event once on that note. Do not search again.
  2. If memory_target is empty, search fix-patterns/ once with one to three root-cause, symbol, or symptom terms.
  3. Update one high-confidence root-cause match. If none exists, create one note. Keep ambiguous roots separate instead of merging by project/customer name.
  4. Stop after the write. Do not run full validate, migration, domain audit, or cross-project candidates during routine recording.
python -X utf8 "$env:USERPROFILE\.codex\skills\obsidian-fix-pattern-memory\scripts\fix_memory.py" upsert --note <canonical.md> --repo . --bug <id> --implementation applied --verification <level> --write

For a new root cause, include title, slug, keywords, scope, symptoms, root cause, key files/functions, fix, verification method, and explicit --domain asr|esp32|none. The script requires the knowledge fields before it writes a new note.

Lookup Path

For explicit memory lookup, regression, similar issue, or cross-branch work, search only fix-patterns/ with one to three precise terms and read at most three genuine matches. A high-confidence match needs compatible symptoms/root cause or code symbols; project, branch, device, or customer text alone is insufficient.

读取记忆 or 根据记忆 authorizes lookup only. A later behavior fix follows the Fast Record Path.

Evidence Event Path

Update only the exact target row:

python -X utf8 "$env:USERPROFILE\.codex\skills\obsidian-fix-pattern-memory\scripts\fix_memory.py" event --note <canonical.md> --repo . --bug <id> --event <event> --evidence "<evidence>" --write

Build/static evidence never becomes device/platform/QA verification. resolved means development-resolved and QA-pending; only closed or equivalent explicit QA evidence supports qa_verified. Reactivation downgrades only the exact target with the materially same symptom.

Validation Debt Reports

For pending device/platform/QA checks or Campaigns, run the read-only report:

python -X utf8 "$env:USERPROFILE\.codex\skills\obsidian-fix-pattern-memory\scripts\validation_debt_report.py" --fix-patterns "$env:USERPROFILE\Documents\Obsidian\CodexVault\Codex\fix-patterns" [filters]

Use project/branch/version/domain filters and --campaign for the requested scope. Read validation debt for grouping, readiness meanings, or open-loop reconciliation. Write a draft/delta only to an explicitly requested standalone path; do not rewrite canonical open-loops. Reporting does not trigger ordinary fix recording or upgrade evidence.

Write Boundary

Record completed behavior-changing code, configuration, or resource fixes. Do not auto-record explanation-only work, comments/formatting, reverted experiments, temporary diagnostics, unrelated environment cleanup, or anything the user says not to record. Never save credentials, full chats, large logs, or private account data.

After a write, report the note path, fix_id, exact target state, evidence level, and whether any other targets remain candidates.

Read fix record schema only when interpreting identity/state transitions. Read maintenance operations only for migration, validation, domain correction, or cross-project candidate audits.

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.