v0.1.0 — initial release

First tagged release. The plugin runs as a Claude Code PreToolUse hook on the Agent tool — when a subagent is about to spawn, declared skills are inlined directly into its prompt, so the agent wakes up with them already in context. No Skill-tool round-trip, no chance of the model "forgetting" to load them.

Highlights

• PreToolUse hook in Python 3 (stdlib only — no dependencies).
• Agent frontmatter declares skills under a single briefing: block:

  briefing:
    skills:
      - perl-core
      - superpowers:brainstorming

  The bare top-level skills: is intentionally ignored — it's reserved for Claude Code itself, so we never collide with future harness keys.
• Resolves bare and plugin:skill namespaced names against project / user / plugin-cache locations.
• Hard-fail on any unresolved skill — permissionDecision: deny with a clear error. No silent drift.
• Ships its own briefing skill that documents authoring patterns (incl. the anti-pattern of restating skills in the agent body when they're already injected) and a 6-step migration recipe.
• 16 unittest cases, CI on Python 3.10 / 3.11 / 3.12.

Install

/plugin marketplace add Getty/briefing
/plugin install briefing@briefing

Distribution

Source archives (.tar.gz / .zip) for this tag are auto-generated by GitHub below. The plugin itself is consumed via git, not as a downloaded artifact, but the archives are useful for offline inspection or air-gapped installs.

See CHANGELOG.md for the full set of changes and PRIVACY.md for the (very short) privacy policy.