health

name	health
description	Run a Brain Check — audit your second brain across Setup, Connections, Beads, Memory, and Inbox. Use when user says "brain check", "health check", "system status", "what's broken", "what needs attention", "check connections", "what's backed up".
argument-hint	["check | setup | connections | beads | memory | inbox | self-test"]
allowed-tools	Bash(python3:*), Bash(node:*), Bash(curl:*), Bash(braynee-self-test:*)

Health Skill — Brain Check

Audits the Braynee system across five areas and surfaces what's broken or backed up right now.

Commands

# Full Brain Check (all five areas)
python3 {baseDir}/scripts/health.py check

# Check tools and plugin scripts are installed
python3 {baseDir}/scripts/health.py setup

# Verify live integrations are reachable
python3 {baseDir}/scripts/health.py connections

# Beads repo health — `bd doctor` + secret/gitignore drift, plus traceability
# hygiene (missing sections, open-but-landed, stale, create-time guard)
python3 {baseDir}/scripts/health.py beads

# Check Claude's context is current
python3 {baseDir}/scripts/health.py memory

# Surface what's backed up or unprocessed
python3 {baseDir}/scripts/health.py inbox

# Plugin self-test — validates every hook, monitor, script, skill, agent
node {baseDir}/../../bin/braynee-self-test

Self-Test (self-test subcommand)

Run when something feels broken with braynee itself, or after editing the plugin source. Validates:

• All 21 hook scripts parse and execute with mock stdin (no crashes)
• hooks.json + monitors.json + plugin.json schema valid
• All 17 skills + 6 agents have valid frontmatter
• All bundled scripts (qmd-wrapper, vault-query, beads-dashboard) are callable
• All 3 monitors can boot

Exit code 0 = everything passed. Non-zero = at least one check failed (output shows which).

For machine-readable output use --json:

node {baseDir}/../../bin/braynee-self-test --json

The same self-test runs in CI on every push to main across Ubuntu, macOS, and Windows.

Brain Check Framework

Setup — Are the tools and scripts installed and working?

• obsidian, bd, node, python3, qmd all available?
• Plugin scripts reachable via {baseDir}?

Connections — Are live integrations actually reachable?

• Beads (issue tracking)
• ProtonMail CLI
• QMD index

Beads — Is the repo's issue tracker healthy, traceable, and free of committed secrets? (runs bd doctor in the current repo)

• No tracked secrets — e.g. a committed .beads/.beads-credential-key (AES-256 key)
• .gitignore carries the beads-required patterns (no drift)
• Surfaces secret/gitignore findings loudly before any repo goes public
• Traceability hygiene — issues missing required sections, implemented-but-still-open (bd orphans), stale issues, and whether the create-time validation guard (validation.on-create) is on. The same checks the beads-auditor agent runs.

Memory — Is Claude loaded with current context?

• ~/.claude/CLAUDE.md reflects current setup?
• Vault CLAUDE.md up to date?
• Priorities reflecting actual focus?

Inbox — What's backed up or unprocessed right now?

• Inbox items waiting in Inbox/
• Sessions not yet synced to vault
• Stale open tasks