agentleFS
revealui-studiorevealui-studio/revskills/skills/revealui-doctor/SKILL.md

revealui-doctor

Skill2 starsChanged 40 days agoMITView on GitHub

Health check for RevealFleet Studio workflow across equal adapters (Claude + Grok at minimum). Verifies hook/rules homes, skill preconditions, git integrity, workboard freshness, daemon status, MCP configs, env leaks, toolchain, and disaster-recovery snapshot state.

  • Reads credentials

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    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:

    Run a health check on Studio-native workflow surfaces for equal adapters.

    line 12
  2. Distinctive:

    If Claude home is missing: do not fail the whole doctor — continue with Grok/other and mark Claude sections SKIP.

    line 30
  3. Very specific:

    Verify these exist and contain .md files when the home is present:

    line 41
  4. Very specific:

    Claude: if ~/.claude/config.json exists, extract mcpServers and verify command / cwd / absolute args paths.

    line 88
  5. Very specific:

    Grok: if ~/.grok/config.toml has MCP / server entries (or document SKIP when config shape is opaque), note presence.

    line 89

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 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

Run a health check on Studio-native workflow surfaces for equal adapters. Report pass/fail (or SKIP with reason) per section. Claude is one adapter; Grok is first-class.

Load helpers:

. "$REVEALFLEET_ROOT/revskills/scripts/lib/session-state.sh"

0. Adapter homes present

Report which homes exist:

AdapterPaths to note
Claude~/.claude/{hooks,rules,commands,settings.json}
Grok~/.grok/{hooks,rules,skills,config.toml,active_sessions.json}
Cursor~/.cursor or project .cursor/ (SKIP if absent)
OpenCodetool-specific (SKIP if absent)

If Claude home is missing: do not fail the whole doctor — continue with Grok/other and mark Claude sections SKIP.

1. Hook syntax

Claude: node --check on every ~/.claude/hooks/*.js when the dir exists. List failures. Confirm settings.json registers PreCompact → snapshot-before-compact.js (last-ditch mechanical capture).
Grok: list ~/.grok/hooks/*.json when present; confirm JSON parses (python3 -m json.tool or jq). FAIL on invalid JSON. Required for snapshot-before-compact: pre-compact.json and stop-snapshot.json (or equivalent PreCompact + Stop entries) pointing at snapshot-before-compact.js. WARN if either event is missing — Grok discards UserPromptSubmit stdout, so without these hooks checkpoints lose pre-compact fidelity. Required for the output cap: cap-tool-output.json with a PostToolUse entry pointing at cap-tool-output.js. FAIL if it is missing.

Compact sync: the authored numbers are revealui/packages/harnesses/src/token-budget.ts, materialized to .revealui/adapters/grok/token-budget.json and copied to ~/.local/share/revealui/hooks/token-budget.json. FAIL if ~/.grok/config.toml disagrees with that JSON: each models[] entry's compaction_at_tokens must equal compactionAtTokens, and auto_compact_threshold_percent must equal autoCompactThresholdPercent (derived from tokens / window, not stored as a second constant). FAIL if cap-tool-output.json is missing. A gate still described as 85% minus 25 with a floor of 50 is stale. Do not compare this budget to Claude context-thresholds.json (that file is a pressure score).

2. Rules directories

Verify these exist and contain .md files when the home is present:

  • $REVEALFLEET_ROOT/revealui/.claude/rules/
  • $JV_REPO/.claude/rules/
  • ~/.claude/rules/ (Claude adapter)
  • ~/.grok/rules/ (Grok adapter; pointer files OK)

3. Skills self-test

Prefer SoT + multi-home:

bash "$REVEALFLEET_ROOT/revskills/scripts/lint-all-skills.sh"

Additionally, for each present command/skill root (~/.claude/commands, ~/.grok skill paths from config if readable):

  • Extract referenced script paths under ~/.claude, $REVEALFLEET_ROOT/revskills, node/bash invocations.
  • Assert each referenced script exists. Report missing.
  • Flag any ~/suite/ references as stale (retired 2026-05-07 → $REVEALFLEET_ROOT/).

4. Git integrity (RevealFleet repos)

For each repo in $REVEALFLEET_ROOT/revealui and $JV_REPO:

cd "$repo" && git fsck --full 2>&1 | grep -E '^(error|fatal|missing)'
ss_empty_objects "$repo"

Empty objects = WSL crash damage. Report loudly.

5. Workboard freshness

Parse $WORKBOARD (neutral $JV_REPO/.revealui/workboard.md; if that file is absent, the Claude adapter render $JV_REPO/.claude/workboard.md). In ## Log, flag [CRASHED] entries older than 24h. The adapter file and ~/.claude/rules/ are not the policy home. Write SSOT for board fragments is $JV_REPO/.revealui/workboard.d.

6. Events log size

$REVEALFLEET_ROOT/revealui/.claude/events.jsonl and $JV_REPO/.claude/events.jsonl. Warn if over 100KB.

7. Daemon + Studio surface health

ss_daemon_alive && echo "daemon: up" || echo "daemon: down"
ss_revvault_alive && echo "revvault: up" || echo "revvault: down"

8. MCP servers

Claude: if ~/.claude/config.json exists, extract mcpServers and verify command / cwd / absolute args paths.
Grok: if ~/.grok/config.toml has MCP / server entries (or document SKIP when config shape is opaque), note presence; do not invent a second MCP schema.
Flag any stale ~/projects/RevealUI path.

9. Git-tracked env files

In both RevealFleet repos: git ls-files '*.env*'. SAFE vs REVIEW classification as before (allowlist from repo security gate; do not hardcode).

10. Settings JSON validity

Claude: validate ~/.claude/settings.json and settings.local.json if present.
Grok: validate ~/.grok/config.toml parses if present (toml via python or skip with note).

11. Toolchain

pnpm -v; node -v; biome --version 2>/dev/null || echo "biome via pnpm exec"
cd "$REVEALFLEET_ROOT/revealui" && test -f flake.lock && nix flake metadata --json >/dev/null 2>&1 && echo "flake: ok" || echo "flake: check"
cd "$REVEALFLEET_ROOT/revealui" && direnv status 2>&1 | tail -3

12. Disaster recovery (WSL snapshot)

Per-repo LTS sync is retired. Flag leftover .claude/lts-mode under $REVEALFLEET_ROOT/*/. Snapshot freshness is Windows-side — SKIP with pointer to revkit check-backup-staleness.ps1 when LTS drive unreachable.

13. Session id auto-resolve (GAP-469)

ss_session_id && echo "session-id: ok" || echo "session-id: unresolved (non-fatal outside a harness session)"

Output

Traffic-light summary (PASS/WARN/FAIL/SKIP) per section and per adapter. JSON sidecar at /tmp/revealui-doctor-last.json (prefer this over Claude-only path names).

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.