agentleFS
MCamnerMCamner/mq-image-analyze/CLAUDE.md

mq-image-analyze

CLAUDE.md1 starsChanged 22 days agoMITView on GitHub

This repo is part of the MQ stack. These instructions add MQ memory read-order rules. They do not replace repo-specific build, test, safety, or release instructions. Default local vault path: $MQ_OBSIDIAN_DIR If MQ_OBSIDIAN_DIR is set, prefer that value. For work related to mq-image-analyze: Stop reading as soon as the task is grounded. mqobsidian is durable memory, not live runtime truth. If the task depends on current code behavior, tests, contracts, CLI behavior, or runtime state, verify in this repo before…

The file

Show its rules and text

We split this CLAUDE.md into the rules it gives an agent

15 rules. About 0 are common, 11 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈11

    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:

    Default local vault path:

    line 10
  2. Very specific:

    If MQ_OBSIDIAN_DIR is set, prefer that value.

    line 14
  3. Very specific:

    Read .mq/context/task-pack.md if it exists and matches the task.

    line 20
  4. Very specific:

    Read $MQ_OBSIDIAN_DIR/memory/learn/agent/mq-image-analyze.md if it exists.

    line 21
  5. Distinctive:

    Read individual pattern notes only if the compressed notes are insufficient.

    line 25
  6. Distinctive:

    Stop reading as soon as the task is grounded.

    line 27
  7. Distinctive:

    Prefer task packs and agent views over full notes.

    line 31
  8. Distinctive:

    Prefer hot/index over pattern notes.

    line 32
  9. Distinctive:

    Do not scan the whole vault by default.

    line 33
  10. Distinctive:

    Do not open multiple pattern notes unless clearly needed.

    line 34
  11. Distinctive:

    If the task depends on current code behavior, tests, contracts, CLI behavior, or runtime state, verify in this repo before making claims.

    line 41
  12. Distinctive:

    When creating notes, summaries, or exports:

    line 46
  13. Distinctive:

    prefer links over duplicated prose

    line 51
  14. Distinctive:

    Do not store or copy secrets, tokens, internal hostnames, raw enterprise logs, or machine-specific private paths.

    line 54
  15. Very specific:

    If mqobsidian is missing, stale, or too weak for the task, say so and verify in the repo.

    line 59

≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md 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
CLAUDE.md

Claude Instructions For mq-image-analyze

This repo is part of the MQ stack.

These instructions add MQ memory read-order rules. They do not replace repo-specific build, test, safety, or release instructions.

mqobsidian Location

Default local vault path:

$MQ_OBSIDIAN_DIR

If MQ_OBSIDIAN_DIR is set, prefer that value.

Read Order

For work related to mq-image-analyze:

  1. Read .mq/context/task-pack.md if it exists and matches the task.
  2. Read $MQ_OBSIDIAN_DIR/memory/learn/agent/mq-image-analyze.md if it exists.
  3. Read $MQ_OBSIDIAN_DIR/systems/mq-image-analyze/hot.md if it exists.
  4. Read $MQ_OBSIDIAN_DIR/systems/mq-image-analyze/index.md if it exists.
  5. Read $MQ_OBSIDIAN_DIR/memory/learn/repos/mq-image-analyze.md if it exists.
  6. Read individual pattern notes only if the compressed notes are insufficient.

Stop reading as soon as the task is grounded.

Low-Token Rules

  • Prefer task packs and agent views over full notes.
  • Prefer hot/index over pattern notes.
  • Do not scan the whole vault by default.
  • Do not open multiple pattern notes unless clearly needed.
  • Summarize instead of replaying long note bodies.

Source-Of-Truth Rule

mqobsidian is durable memory, not live runtime truth.

If the task depends on current code behavior, tests, contracts, CLI behavior, or runtime state, verify in this repo before making claims.

Writing Rules

When creating notes, summaries, or exports:

  • separate facts, interpretation, and recommendation
  • keep outputs compact
  • preserve timestamps and provenance when relevant
  • prefer links over duplicated prose
  • avoid raw dumps

Do not store or copy secrets, tokens, internal hostnames, raw enterprise logs, or machine-specific private paths.

Fallback Rule

If mqobsidian is missing, stale, or too weak for the task, say so and verify in the repo. Do not invent continuity.

GOVERNOR — mansys/mcamner

<!-- Skriven av Fable 5, 2026-07-07. Klistras in i ~/.claude/CLAUDE.md eller per repo. -->

Kommunikation

  • Svara på svenska om inget annat sägs. Kort och direkt. Ingen hype, inga superlativ, inga "Great question!".
  • Ärlig bedömning före artighet. Säg "det här är en dålig idé" när det är det, med skäl.
  • Osäkerhet: säg "kan inte bekräfta" istället för att gissa. Hitta aldrig på källor, siffror eller API:er.
  • Publik text (LinkedIn, README): faktisk precision, ingen marknadsföringston. Skriv som en människa, inte som en AI.

Kod

  • Kirurgiska ändringar. Rör inte kod utanför uppgiften. Ingen "passade på att refaktorera".
  • Inga onödiga abstraktioner. Enklaste lösning som håller.
  • Redovisa antaganden explicit innan implementation. Definiera verifierbart framgångskriterium.
  • Verifiera innan du deklarerar klart: kör testet, läs outputen, visa beviset.
  • TDD vid features och bugfixar där det är rimligt.
  • Läs faktiska filer i repot istället för att gissa struktur. Fråga inte "vill du att jag tittar?" — titta.

Arbetssätt

  • Kör vidare på självklara nästa steg utan att fråga. Fråga endast vid destruktiva operationer (delete, force-push, prod).
  • Vid felsökning: reproducera → isolera → diagnostisera → fixa. Inte "prova det här och se".
  • En fråga i taget om förtydligande behövs, och bara om svaret inte redan finns i kontexten.

Miljö

  • macOS: MQ-stacken. mq-mcp är MCP-servern (hal_repo_report, read_repo_file, run_mqlaunch_doctor, record_learning). Bridget är agenten (bridge.py, bridget_context.py). Repon: mq-mcp, mq-agent, mq-hal, macos-scripts, mqobsidian.
  • Windows (Region Stockholm): PowerShell-svit med mongo-prefix (mongostart, mongoApps, mongoSys, mongoKommand, loggMongo). GPO-begränsad miljö — räkna med workarounds.
  • Fedora-testmaskin (Dell Latitude 5290): Fish/bash.
  • IT-domän: IGEL OS 12, UMS, Citrix CVAD, Intune/Entra ID. Svensk offentlig sektor/vård — säkerhet och spårbarhet väger tungt.

Konventioner

  • Namnprefix: mq- (macOS), mongo- (Windows).
  • Estetik: JetBrains Mono, amber/dark terminal, HAL 9000/Amiga-tema.
  • Dokumentation ofta bilingualt SV/EN.

Effort (Claude Code)

  • ultrathink: enskilt svårt problem, en tur. Verifierat nyckelord.
  • ultracode: sessionsinställning, xhigh + dynamic workflows. Endast för stora parallella jobb — tokenkostnaden är öppen. Kräver xhigh-kapabel modell (Fable 5, Opus 4.8/4.7).
  • Vardagsläge: /effort high.

Claude Code — hävstång

  • Använd skills proaktivt när de matchar uppgiften och är tillgängliga: /verify före commit av icke-trivial kod, /code-review för buggjakt, /debug vid svårfångade fel. Nämn aldrig en skill utan att faktiskt köra den.
  • Bred sökning: kör oberoende verktygsanrop parallellt och delegera fan-out till Explore-subagenten när den är installerad/tillgänglig, i stället för seriell grep/read.
  • "Klart" kräver verifiering i verkligheten — driv det faktiska flödet och läs outputen, inte bara typecheck/tester.
  • Efter buggfix i MQ-repon: spara det icke-uppenbara (API-kontrakt, konventioner) i auto-minne och via record_learning när mq-mcp/verktyget finns — selektivt, bara det som påverkar framtida arbete.
  • CodeGraph först (se ovan) i indexerade repon innan grep/read.

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.