Claude Skills by antoinecellerier

Validates a change to the audio output path against measured on-device ground truth. Use after any change to FIR generation, gain staging, filter parameters, or the EasyEffects / PipeWire output chain, before adopting or shipping it — and whenever the user asks to "test on device", "capture the EE response", "compare against DAX", or confirm a change didn't regress audibly. This skill gates on an audio handoff and drives the live-EE capture → DAX/EE compare → listening route end-to-end. Do NO...

Audits the instruction files (CLAUDE.md, .claude/rules, .claude/skills, tools/measure_dax/CLAUDE_WINDOWS.md) for accuracy and bloat against .claude/rules/instructions.md, and proposes a concrete edit list. Runs before committing a change to any of them, and periodically as a maintenance pass. It verifies every reference still resolves, flags references to gitignored file paths, checks line and description budgets, finds duplication across CLAUDE.md / rules / skills / docs / memory, and scans ...

Audits the user-facing terminal copy changed over a git range for factual truth rather than readability, by fanning out reviewers partitioned by evidence source and triaging what survives. Use before a release, after a batch of messaging work, or when the user asks to "check the messages are accurate", "verify what we're telling people", "audit the output for correctness", or doubts a claim a run prints. Complements /user-review, which asks whether a first-time reader can act on a message; th...

Reviews the user-facing docs — README.md and the user guides under docs/README.md "Using it" — by running them past subagent reviewers role-playing fixed personas: a first-time visitor, a user troubleshooting a symptom, and a PipeWire-only user. Reports severity-ranked findings after triage. Use after changing README or a user guide, before a release, and whenever the user asks for a "persona review", to "check the docs read well", or whether a newcomer can get from zero to working sound. Com...

Guides triaging GitHub issues and drafting or posting replies in this repo. Load it when starting to triage or investigate an issue ("check issue #NN", a new device report, a bug report), including in plan mode, because it shapes what the investigation must produce. Load it again before drafting any issue or PR reply or running `gh issue comment` / `gh pr comment`. Covers structure and tone, what may be asserted versus framed as a hypothesis, citation and link rules, runnable experiments, and...

Guides triaging a kernel-sound-watch hit comment on the "Kernel sound-tree watch" issue (#40) — the weekly workflow's per-tag report of `.github/kernel-watchlist.txt` grep hits against a new tiwai/sound pull tag. Load this when asked to "analyse/check the latest kernel watchlist hit", when a kernel-sound-watch comment needs a verdict, or before appending a `### Triage` section to one. Covers reading the commit range cheaply, judging a hit against the watch that fired it, verifying claims agai...

Reviews the scripts' user-facing terminal output by running it past subagent reviewers role-playing a first-time user, then reports severity-ranked findings. Use after changing any message a user reads — end-of-run blocks, warnings, flag menus, asks, phase banners — and whenever the user asks to "review the output", "check how this reads", "run it past a user", or wants to know whether a message is understandable and actionable. The test suite traps structure; this catches copy that is confus...