layerbrain

@layerbrain/skills

Skills that teach coding agents how to drive the brain CLI.

Quick start

npm i -g @layerbrain/brain
brain login

npx @layerbrain/skills install

Re-run any time to upgrade.

Supported targets

Target	File
Claude Code	.claude/skills/<name>/SKILL.md
Cursor	.cursor/rules/<name>.mdc
Cline	.clinerules (managed block)
AGENTS.md	AGENTS.md (managed block)

Pass --agent <id> to force one, or let autodetect handle it.

What gets installed

Skill	What the agent learns
brain-machines	Provision VMs, snapshot, restore, SSH, pick regions
brain-storage	Manage backends, buckets, presigned URLs
brain-networks	Networks, ACL rules, traffic flow debugging
brain-secrets	Create, rotate, reveal secrets
brain-iam	Accounts, API keys, organizations, memberships
brain-webhooks	Create, rotate signing secrets, manage webhooks
brain-billing	Read statements and subscriptions

Plus a shared references/auth.md codifying the credential model (~/.layerbrain/credentials.toml or LAYERBRAIN_API_KEY).

Commands

npx @layerbrain/skills install                                # all skills, autodetect target
npx @layerbrain/skills install brain-machines brain-storage   # subset
npx @layerbrain/skills install --agent claude-code            # force one target
npx @layerbrain/skills install --agent claude-code --scope user
npx @layerbrain/skills list                                   # what's installed
npx @layerbrain/skills remove brain-billing                   # remove one
npx @layerbrain/skills detect                                 # show detected targets
npx @layerbrain/skills agents                                 # list every supported target ID
npx @layerbrain/skills skills                                 # list available skills
npx @layerbrain/skills help

Append-style targets (AGENTS.md, .clinerules) use <!-- @layerbrain/skills:start --> / :end markers, so re-running upgrades cleanly without touching your other rules.

Authentication

Skills never authenticate the agent — that's a human-only step. Run brain login once. Credentials live at ~/.layerbrain/credentials.toml. For headless / CI use, set LAYERBRAIN_API_KEY.

Every skill preamble tells the agent to run brain whoami first and stop if it fails.

License

MIT