health-check

name	health-check
description	Verify CLI tool availability, authentication status, config syntax, MCP connectivity, and symlink integrity for the Manifest environment.

Health Check Skill

Run a comprehensive diagnostic of the Manifest agent environment to detect misconfiguration, missing tools, broken symlinks, or authentication issues.

Scope note: This skill covers config syntax, MCP connectivity, symlinks, labels, and script executability. For a quick terminal check of parallel orchestration readiness (enabled agents, state directories), run: ~/.claude/scripts/check_status.sh (also available as parallel_agent.py --status).

Checks

Execute each check category below. Collect results into a summary table.

1. CLI Tool Availability

For each service in .claude/config/services.yml where enabled: true:

command -v <tool_command> &>/dev/null
<tool_command> --version 2>/dev/null || <tool_command> version 2>/dev/null

Report: installed (with version) or missing.

2. Authentication Status

Check authentication for each enabled service:

• Claude: claude auth status 2>/dev/null or check ANTHROPIC_API_KEY is set
• Gemini: gemini auth status 2>/dev/null or check GOOGLE_API_KEY is set
• Cursor: cursor --version (no separate auth check; presence implies configured)
• GitHub CLI: gh auth status 2>/dev/null
• GitLab CLI: glab auth status 2>/dev/null

Report: authenticated, unauthenticated, or not applicable.

3. Configuration Syntax

Validate YAML and JSON config files:

python3 -c "import yaml; yaml.safe_load(open('<file>'))" 2>&1
python3 -c "import json; json.load(open('<file>'))" 2>&1

Files to check:

• .claude/config/command_config.yml
• .claude/config/parallel_agent.yml
• .claude/config/services.yml
• .claude/config/validation_criteria.yml
• .claude/config/linear_triage.yml
• .claude/config/mcp_servers.yml
• .claude/config/labels.yml
• .claude/settings.local.json

Report: valid or error details.

4. MCP Server Connectivity

For each server in .claude/config/mcp_servers.yml:

curl -s -o /dev/null -w "%{http_code}" --max-time 5 <url>

Report: reachable (HTTP status) or unreachable.

5. Symlink Integrity

Verify all cross-platform symlinks are intact:

# Expected symlinks (target → source)
.cursor/scripts  → ../.claude/scripts
.cursor/config   → ../.claude/config
.cursor/prompts  → ../.claude/prompts
.cursor/skills   → ../.claude/skills
.cursor/.plans   → ../.claude/.plans
.gemini/scripts  → ../.claude/scripts
.gemini/config   → ../.claude/config
.gemini/prompts  → ../.claude/prompts
.gemini/skills   → ../.claude/skills
.gemini/.plans   → ../.claude/.plans
.codex/scripts   → ../.claude/scripts
.codex/config    → ../.claude/config
.codex/prompts   → ../.claude/prompts
.codex/skills    → ../.claude/skills
.codex/.plans    → ../.claude/.plans
.antigravity/scripts → ../.claude/scripts
.antigravity/config  → ../.claude/config
.antigravity/prompts → ../.claude/prompts
.antigravity/skills  → ../.claude/skills
.antigravity/.plans  → ../.claude/.plans

Only check symlinks for services marked enabled: true in .claude/config/services.yml (e.g. skip .cursor/.codex when disabled).

For each: check if symlink exists and target is valid.

Report: intact, broken (dangling), or missing.

6. Label Registry Validation

Verify the label registry is valid and labels are consistent:

1. Check .claude/config/labels.yml exists and is valid YAML
2. Verify each label has required fields: name, color, description, platforms
3. Verify color values are valid 6-digit hex (no leading #)
4. Check for duplicate label names
5. If gh is available, run gh label list --json name,color and compare against registry

# Validate labels.yml syntax
python3 -c "
import yaml, sys
with open('.claude/config/labels.yml') as f:
    data = yaml.safe_load(f)
labels = data.get('labels', [])
for label in labels:
    assert 'name' in label, f'Missing name in label: {label}'
    assert 'color' in label, f'Missing color in {label[\"name\"]}'
    assert 'platforms' in label, f'Missing platforms in {label[\"name\"]}'
    assert len(label['color']) == 6, f'Invalid color hex in {label[\"name\"]}: {label[\"color\"]}'
print(f'{len(labels)} labels validated')
"

Report: valid (N labels) or error details.

7. Browser-Use Availability (Info)

Check if browser-use is available for E2E testing:

# Check CLI
command -v browser-use &>/dev/null && browser-use --version 2>/dev/null

# Check Python module
python3 -c "import browser_use; print(browser_use.__version__)" 2>/dev/null

Report: installed (with version) or not installed.

This check is informational only — browser-use is an optional tool. Report as info (not fail) when missing.

8. Script Executability

Verify all scripts in .claude/scripts/ are executable:

[[ -x "$script" ]]

Report: executable or not executable.

SkillClaw (if enabled)

Only when skillclaw.enabled: true in ~/.claude/config/services.yml:

# Daemon health (fail-open: a red here means capture is off, not that agents are broken)
curl -sf --max-time 0.3 http://127.0.0.1:8765/health && echo "daemon: up" || echo "daemon: down"

# Wrapper functions present in the shell profile
grep -q "MANIFEST SKILLCLAW WRAPPERS" "${ZDOTDIR:-$HOME}/.zshrc" && echo "wrappers: present" || echo "wrappers: MISSING"

# Storage locked down (must be 700) — GNU-first so it works on Linux and macOS
stat -c '%a' ~/.skillclaw 2>/dev/null || stat -f '%Lp' ~/.skillclaw

Report daemon: down as INFO (capture paused, agents unaffected), but wrappers: MISSING or storage perms != 700 as WARN.

Output Format

## Health Check Report

| Category | Check | Status | Details |
|----------|-------|--------|---------|
| CLI Tools | claude | pass | v4.x.x |
| CLI Tools | gemini | pass | v1.x.x |
| CLI Tools | cursor | fail | Not installed |
| Auth | claude | pass | Authenticated |
| Auth | gh | warn | Not configured |
| Config | command_config.yml | pass | Valid YAML |
| Config | services.yml | fail | Parse error on line 12 |
| MCP | sentry | pass | HTTP 200 |
| MCP | linear | warn | HTTP 401 (auth required) |
| Symlinks | .cursor/scripts | pass | Intact |
| Symlinks | .gemini/config | fail | Broken |
| Labels | labels.yml | pass | 5 labels validated |
| Scripts | parallel_agent.py | pass | Executable |

### Summary

- pass: N checks passed
- warn: N warnings (non-blocking)
- fail: N failures (action needed)

Tool Usage

• Bash: Run CLI version checks, auth status, curl, stat, test
• Read: Read config files for validation
• Glob: Find config and script files
• Grep: Search for configuration values