<!-- Generated by scripts/sync-mirrors.sh from the canonical sources — do not edit. -->

# Repository Instructions

Assistant configuration is canonical under `.claude/`. Codex discovers this
repository's skills from `.agents/skills/` (generated). Regenerate every
surface with `scripts/sync-mirrors.sh`; verify with `--check`.

## Skills and commands

Codex discovers both from `.agents/skills/`.

Reusable skills (20): accent-dial, audit-references, bake-off, cold-review, critic-apply, critic-panel, filter-tells, humanize, inject-vernacular, match-outline, match-structure, match-voice, patent-disclosure, pattern-language, reverse-outline, stitch-flow, tighten-style, tune-anchors, update-references, voice-critic.


Python-backed skills run in a pixi environment that ships beside them:
run `.agents/scripts/ensure-env.sh` once per machine, then invoke scripts
with `pixi run --manifest-path .agents/pixi.toml python <script>`.
API credentials resolve through the repository's gitignored `.secrets/`
directory, never from the skills tree — see the secrets rule below.

## Conventions

These are summaries; the canonical rule files hold the detail.

- **The `.secrets/` Directory Contract** — API credentials belong to the repository you are working in, never to the skills. See `.claude/rules/secrets.md`.
- **Technical Document Types** — We catalog the document types of standards bodies and engineering practice. See `.claude/rules/technical-document-types.md`.
- **The `writing-voice/` Directory Contract** — Any writing repository may carry a `writing-voice/` directory of exemplar samples that define the target voice for text generated in that repository. See `.claude/rules/writing-voice.md`.

Before committing, run the repository's consistency check if it defines
one, and commit after each round of edits rather than accumulating them.
