agentleFS
egregore-labsegregore-labs/egregore/.claude/skills/checkup/SKILL.md

checkup

Skill293 starsChanged 10 days agoMITView on GitHub

Run diagnostics on an Egregore environment and render a health report, repairing safe local Runtime problems when possible. Use for /checkup, a health check, or when something is broken and the root cause is unclear.

  • Reads credentials

7 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use report.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • MicrosoftSkill

    report

    Turn an exploration (threads, findings, charts) into a single Markdown report — note, blog post, executive summary, KPI dashboard, slide brief, or multi-section analytical report, with embedded charts.

    Also in 3 projects that use this

The file

Show its rules and text

We split this skill into the rules it gives an agent

14 rules. 0 are common, 8 are distinctive, and 6 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 8

    Distinctive

    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:

    Run diagnostics on an Egregore environment and render a health report, repairing safe local Runtime problems when possible.

    line 15
  2. Very specific:

    Use the printed value as {mode}.

    line 25
  3. Distinctive:

    Run three sequential batches.

    line 28
  4. Distinctive:

    Never run Connected network probes beside local checks.

    line 34
  5. Distinctive:

    Read the one typed Runtime health surface as JSON:

    line 39
  6. Very specific:

    Do not invoke qmd, inspect QMD process tables, guess ports, or read another instance's runtime directory.

    line 46
  7. Distinctive:

    If the Runtime worker or indexes are unavailable while canonical memory is ready, run the safe Runtime lifecycle adapter once:

    line 69
  8. Distinctive:

    Do not auto-replace API keys, repair graph state, publish, or send a notification.

    line 97
  9. Very specific:

    Never mutate Git, canonical memory, or framework files during diagnosis.

    line 112
  10. Very specific:

    Use ✓, ⚠, and ✗ for pass, degraded, and unavailable.

    line 146
  11. Distinctive:

    Never expose raw JSON, tokens, environment values, or unformatted probes.

    line 152
  12. Distinctive:

    Never use Graph or a control-plane response as canonical health authority.

    line 153
  13. Very specific:

    Never call QMD directly; instance ownership belongs to Egregore Runtime.

    line 154
  14. Distinctive:

    Never send, publish, commit, pull, or push as part of checkup.

    line 155

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “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

Egregore checkup

Diagnose the Egregore Runtime first. Treat canonical Markdown and its instance-owned Local Retriever as the product baseline. Graph, control-plane, and notification probes are explicit Connected integrations; their failure must never make canonical Local Runtime state look unavailable.

When to invoke

Run diagnostics on an Egregore environment and render a health report, repairing safe local Runtime problems when possible. Use for /checkup, a health check, or when something is broken and the root cause is unclear.

Procedure

Detect mode without attempting a network call:

bash bin/config-get.sh mode

Use the printed value as {mode}; quote it in single quotes when you use it in a command, and write any single quote inside the value as '\''. If it prints local, skip Connected integrations; if it prints connected, include them after the local checks.

Run three sequential batches. Parallelize only within a batch.

  1. Runtime and local configuration.
  2. Connected integrations, only when {mode} is connected.
  3. Workspace and Git checks.

Never run Connected network probes beside local checks; a timeout must not cancel or obscure the Runtime result.

Batch 1: canonical Runtime

Read the one typed Runtime health surface as JSON:

EGREGORE_ROOT="$PWD" PYTHONSAFEPATH=1 PYTHONPATH="$PWD${PYTHONPATH:+:$PYTHONPATH}" \
  python3 -m egregore_runtime.harness_cli status --json

Do not invoke qmd, inspect QMD process tables, guess ports, or read another instance's runtime directory. The adapter owns those mechanics. Interpret the returned fields as follows:

  • canonical_state_ready — foundational. Fail only when canonical memory is unavailable.
  • adapter=qmd-local, collection, and index_path — the resolved instance-owned Local Retriever. Display these without exposing another instance or deriving ownership yourself.
  • runtime_state, runtime_pid, and runtime_endpoint — owned worker readiness. semantic_ready is healthy; semantic_index_building is a warning with lexical fallback; runtime_unavailable is repairable.
  • bm25_ready — local keyword readiness. Warn when false.
  • semantic_ready and embedding_state — semantic readiness. A warming or missing model is degraded, not canonical data loss.
  • index_spec_version, index_revision, and source_revision — freshness. Report drift; never repair it by editing index files.
  • warnings — show concise, redacted messages.

Also validate egregore.json, the canonical memory link/repository, and the mode-appropriate key names in .env. Never print secret values. Local mode requires no control-plane key.

If the Runtime worker or indexes are unavailable while canonical memory is ready, run the safe Runtime lifecycle adapter once:

bash bin/search.sh start

Then re-read Runtime health once. This starts or reuses only the worker owned by this Egregore instance and schedules semantic warming behind the adapter.

Batch 2: Connected integrations

Skip this batch completely in local mode. In connected mode, label every row as an integration, keep it outside the foundational Runtime verdict, and run:

timeout 10 bash bin/graph-projection.sh verify --enable 2>&1; echo "EXIT:$?"
timeout 10 bash bin/notification.sh status 2>&1; echo "EXIT:$?"
  • Hosted relationship index — pass when the gateway accepts the configured credentials and the projection is reachable. Offline, auth, or stale projection results are Connected warnings; canonical Markdown remains authoritative.
  • Notification adapter — read the sanitized Runtime status field. Treat ok (connected) or configured (local) as healthy. Telegram is optional and never part of Runtime readiness.

Do not auto-replace API keys, repair graph state, publish, or send a notification. Report the exact integration boundary that failed.

Batch 3: workspace

Check locally:

  • the configured base branch exists and compare it with origin/<base> only when an origin is already configured;
  • the current worktree and memory repository are clean or clearly report their dirty-file counts;
  • the framework version is readable locally. Do not fetch merely to render a health report;
  • the shell launcher points at this instance when a launcher is configured.

Never mutate Git, canonical memory, or framework files during diagnosis. Offer $pull, $update, or $setup as the next Egregore action when relevant; do not tell the user to construct repair commands manually.

Rendering

Render one compact diagnostic box. In connected mode add a separate CONNECTED INTEGRATIONS section after RUNTIME; in local mode omit it.

┌──────────────────────────────────────────────────────────────────────┐
│  ⊕ CHECKUP                                          {date}         │
├──────────────────────────────────────────────────────────────────────┤
│  CONFIG                                                              │
│  ✓ Instance — {org} ({mode})                                        │
│                                                                      │
│  RUNTIME                                                             │
│  ✓ Canonical memory — ready · source {source_revision}              │
│  ✓ Local Retriever — qmd-local {adapter_version} · owned            │
│  ⚠ Semantic index — warming · lexical fallback ready                │
│  ✓ Worker — pid {pid} · {endpoint}                                  │
│                                                                      │
│  CONNECTED INTEGRATIONS                                              │
│  ⚠ Hosted relationship index — offline · local memory ready         │
│  ✓ Telegram — connected                                             │
│                                                                      │
│  WORKSPACE                                                           │
│  ✓ Memory repository — clean                                        │
│  ✓ Git — {base} synced                                               │
├──────────────────────────────────────────────────────────────────────┤
│  Runtime ready · {warnings} warnings · integrations {state}         │
└──────────────────────────────────────────────────────────────────────┘

Use ✓, ⚠, and ✗ for pass, degraded, and unavailable. State the Runtime verdict independently from Connected integration health. After a safe Runtime repair, append only the rechecked result.

Rules

  • Never expose raw JSON, tokens, environment values, or unformatted probes.
  • Never use Graph or a control-plane response as canonical health authority.
  • Never call QMD directly; instance ownership belongs to Egregore Runtime.
  • Never send, publish, commit, pull, or push as part of checkup.
  • Emit telemetry only through the normal command wrapper, best-effort.

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.