debug-voice
An agent skill by cursor, from cursor/plugins. Tags: audio, debugging, logging, troubleshooting, validation.
What it does
Use when the user runs /debug-voice, says voice mode has flaws, asks to see or capture what happened in a Grok realtime voice session, or a voice integration has no debug logging yet. Proposes a plan, then installs a dev-only log pipeline (client logger → local NDJSON) in the app's own language and conventions, then runs the fix loop: match the user's report to log signatures, fix one thing, re-test.
Install
With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:
npx skills add cursor/plugins --skill debug-voice
Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):
git clone --depth 1 https://github.com/cursor/plugins
cp -r plugins/grok-voice/skills/debug-voice ~/.claude/skills/debug-voice
Safety box score
Grade B, low risk: 16 of 100 risk points, rated 2026-10-16. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.
This is a dev-only voice-debugging skill that adds local logging, a local HTTP sink, and UI changes inside the app. The main blast radius is limited to the project directory, but it does create persistent voice-session logs and depends on remote xAI docs for event names.
| Category | Level | Why |
|---|---|---|
| Code & shell execution | 1 (low) | It includes local shell verification commands (`curl`, `jq`) and instructs the agent to install app-specific logging code, but it does not ship an executable script or ask for `eval`/downloaded code. |
| Network access | 1 (low) | The only explicit data flow is to a local dev sink (`POST /api/voice/log` on localhost); the skill also references an external docs URL, but not as a code-download mechanism. |
| Filesystem reach | 1 (low) | It writes `.voice-logs/<sessionId>.ndjson`, asks to add `/.voice-logs` to `.gitignore`, and modifies app files, but the scope is intended to stay inside the project tree. |
| Secrets & credentials | 0 (none) | The skill explicitly says "Never tokens, keys, or raw audio" and the log contract redacts those fields; it does not request credential stores, browser sessions, or pasted secrets. |
| Privilege & persistence | 0 (none) | No sudo, global installs, services, scheduled tasks, or shell/OS settings are requested. |
| Instruction hijack surface | 1 (low) | It tells the agent to consult a remote docs page before adding event names (`"confirm in https://docs.x.ai/..."`), but it still keeps the workflow user-confirmed and does not contain hidden prompt text. |
| Irreversible actions | 0 (none) | It is a debug-and-fix workflow only; it does not deploy, publish, send messages, or make external side effects beyond local logging. |
| Transparency | 0 (none) | The skill text is plain markdown with no binaries, base64 blobs, or generated installers. |
Before installing
- Read the app’s `POST /api/voice/log` path and confirm it is dev-gated (`dev` or `VOICE_LOG=1`).
- Check that `.voice-logs/` stays inside the repo and is added to `.gitignore`.
- Verify the logger redaction rules prevent raw audio, tokens, and long strings from being written.
- Run the smoke `curl` test in a scratch dev session and inspect one `.voice-logs/<id>.ndjson` file before using it on real voice runs.
Rated 2026-10-16 by an automated review of 1 file from the skill's repository.
Source
- Repository
- cursor/plugins (all skills from this repository)
- Path
- grok-voice/skills/debug-voice/SKILL.md
- Branch
- main
- Collection
- cursor
- Updated
- 2026-09-23
Related skills
- fix-ci — Find failing PR checks, inspect logs or external check links, and apply focused fixes
- principle-fix-root-causes — Apply when debugging.
- reproduce-and-fix-issues — Reproduce triaged Slack bugs through a configured app-control adapter, verify existing fixes.
- claude-skills-troubleshooting — Diagnose and resolve Claude Code plugin and skill issues.
- debugging-network-issues — Investigates network, streaming, and protocol failures with falsification-first layered experiments.
- investigate-first — Diagnose ambiguous failures before editing.