Help

/li:help

Meta-skill. Lists what Lintel makes available in this session so the operator knows what they can run, without grepping the plugin tree. A quick in-session lister; for the full generated reference see /li:catalog (it builds skills/CATALOG.md from frontmatter), and for install/version health see /li:doctor.

When to use

• First session after install — discover what's there
• Picking the right skill for a task (filter by category)
• Onboarding a teammate to Lintel
• Debugging "is this skill available?" / "is it the right CLI?"

Inputs

Optional flags:

• --category <name> — filter to one category (plan, qa, ship, compliance, voice, meta, ops)
• --voice <internal|customer|mixed> — filter by voice tier
• --cli <claude-code|codex|copilot> — show only skills supported on a specific CLI
• --verbose — include description per entry (default: one-line entries)

No arguments: full list grouped by category, one line per skill.

Workflow

Lintel ships as a plugin, so skills/agents/hooks are resolved from the installed plugin — not from a hand-managed ~/.claude/ tree.

1. Resolve the plugin root: ${CLAUDE_PLUGIN_ROOT:-$(git rev-parse --show-toplevel 2>/dev/null)}.
2. Glob "$root"/skills/*/SKILL.md for the available skills (they are invoked namespaced as /li:<name>).
3. Glob "$root"/agents/*/*.md for the subagent fleet (resolve by name via the Agent tool).
4. For each skill/agent, parse YAML frontmatter for: name, description, voice, cli_support, color.
5. Filter per operator's flags.
6. Group by category (heuristic from skill name: plan-* → plan, qa* → qa, etc.).
7. Output structured list. Read the plugin version from "$root"/.claude-plugin/plugin.json.

Report format

Default (full list):

Lintel v<version> — <N skills>, <M agents>, <K hooks>

## Plan (<count>)
- /li:plan-ceo-review     [internal, all CLIs] — Strategy & scope review
- /li:plan-eng-review     [internal, all CLIs] — Architecture & tests review
- /li:plan-design-review  [internal, claude-code] — UI/UX gaps
- /li:plan-devex-review   [internal, all CLIs] — DX gaps

## QA + debug (<count>)
- /li:qa                  [internal, claude-code] — Real-browser testing
- /li:qa-only             [internal, claude-code] — Test-only flow
- /li:investigate         [internal, all CLIs] — Bug forensics
- /li:review              [internal, all CLIs] — Diff-scoped pre-ship review

## Voice (<count>)
- /li:eval               [internal, claude-code] — Calibrate the pack's voice corpus

## Compliance (<count>)
- /li:compliance-gate    [internal, all CLIs] — Run the active pack's compliance gates

## Meta + ops (<count>)
- /li:context-save       [internal, claude-code] — Save checkpoint
- /li:context-restore    [internal, claude-code] — Read checkpoint
- /li:clean              [internal, claude-code] — Manual self-maintenance
- /li:help               [internal, all CLIs] — This skill
- /li:doctor             [internal, all CLIs] — Install/version/hook health check

## Agents (the plugin fleet)
- CodeReviewer           [engineering, claude-code] — Reviews diffs for correctness/quality/security
- SecurityAuditor        [security, claude-code] — Injection/secret/auth-bypass audit
- ...  (run /li:catalog for the full list)

## Hooks (activation per ADR-0008)
- Auto-registered on plugin install (9): session-digest, secret-scan-block, customer-data-block,
  no-secrets-in-edit, no-direct-main-push, memory-budget-warn, cycle-incomplete-warn,
  cycle-position-inject, no-customer-data-in-message
- Opt-in module warn-hooks (24): the ta/da/sc/dh/tq-* warn hooks — enable via a `~/.lintel/hooks`
  symlink. Run /li:hooks-status for live state.

Filtered by category:

> /li:help --category qa
QA + debug skills:
- /li:qa            [internal, claude-code]
- /li:qa-only       [internal, claude-code]
- /li:investigate   [internal, all CLIs]
- /li:review        [internal, all CLIs]

Filtered by CLI (Copilot user):

> /li:help --cli copilot
Lintel workflows supported on GitHub Copilot:
- /li:plan-ceo-review   [internal]
- /li:plan-eng-review   [internal]
- /li:investigate       [internal]
- /li:help              [internal]
- /li:doctor            [internal]

Use native li-* skills from .github/skills (for example /li-plan where slash discovery is available), or ask Copilot to load the matching skill. The CLI plugin supplies the same adapters. Lintel hooks are not ported.

Edge cases

• Plugin root not resolvable: report "Lintel plugin not detected — install via /plugin install li@jokerman-lintel, then restart the session."
• Skill missing frontmatter fields: flag the skill (⚠ <name>: missing cli_support) — report it.
• Filter matches zero skills: report "no skills match these filters" + suggest dropping a flag.
• Version drift across CLIs: defer to /li:doctor, which is the cross-CLI drift checker.

Failure modes

• No skills/ under the plugin root: report "no Lintel skills found at the plugin root."
• Frontmatter parse error in a skill: skip that skill, report which one was unparseable.

Examples

> /li:help
[full list output]

> /li:help --category compliance --verbose
Lintel compliance skills:
- /li:compliance-gate  [internal, all CLIs]
    Runs the active pack's compliance gates (`resolve_pack_field
    compliance.hooks`; none by default). Surfaces results; does NOT
    enforce — operator confirms.

> /li:help --voice customer
Lintel customer-voice skills:
- /li:eval  [claude-code]

See also

• /li:catalog — the full generated skill reference (built from frontmatter)
• /li:doctor — install + version + hook-firing health (cross-CLI)
• /li:hooks-status — live hook activation state
• AGENT-INSTRUCTIONS.md — canonical session-start ritual