Use when the user wants to check installed Claude Code plugins, skills, agents, or MCP configs for leaked secrets or suspicious/exfiltration patterns.
Scanned 9/4/2026
Install to Claude Code
npx -y skills add sonomoshq/Canary --skill audit --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Audit?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/sonomoshq-audit)More formats (shields.io, HTML) on the badges page.
---
name: audit
description: Use when the user wants to check installed Claude Code plugins, skills, agents, or MCP configs for leaked secrets or suspicious/exfiltration patterns.
user-invocable: true
disable-model-invocation: true
argument-hint: "[--record|--strict]"
allowed-tools: Bash(bash:*), Bash(cat:*), Bash(jq:*)
---
# Sonomos Extension Audit
Scan every installed skill, agent, plugin, and MCP/settings config for leaked secrets and suspicious exfiltration patterns — the same kind of PII/secret detectors Canary runs on your conversations, turned on the extensions themselves.
Run:
```bash
bash "${CLAUDE_PLUGIN_ROOT}/scripts/audit-plugins.sh" $ARGUMENTS
```
`$ARGUMENTS` may be empty (report-only), `--record` (also append one leaks.jsonl summary line per flagged extension), `--strict` (exit 2 if anything CRITICAL was found), or both.
## After it runs
Summarize the findings for the user in your own words:
- Which extensions were flagged, their score/band (LOW/MEDIUM/HIGH/CRITICAL), and the top few findings for each (rule, severity, file:line when available).
- **Never print a raw secret value** — the script's output is already redacted by `detectors.sh`; just relay what it printed.
- If nothing was flagged, say so plainly — that's a good result.
Then suggest concrete next steps based on what was found:
- Rotate any credential that showed up as a leaked secret (API key, token, private key) — assume it's compromised once it's sat in a plugin/skill file.
- For CRITICAL findings (collector-domain, env-harvesting) or a CRITICAL-band extension, recommend removing or disabling that extension until it's been reviewed line by line.
- For a finding that's actually a documented example or an intentional false positive, point out that adding `# canary-ignore` to that line (or, for `.md` files, putting it in a fenced code block introduced by a line containing the word "example") will suppress it on the next run.
## Notes
- This is a report-only tool: it never modifies the scanned files, and only writes to `leaks.jsonl` when `--record` is passed.
- Canary's own plugin directory is skipped automatically — it legitimately ships the detector pattern library this audit is built on.
Copyright © 2026 Sonomos Inc. All rights reserved.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!