Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/internlm/wildclawbench/canarynpx skills add InternLM/WildClawBench --skill canarygit clone --depth 1 https://github.com/InternLM/WildClawBenchWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/internlm/wildclawbench/canary)<a href="https://agentmods.dev/skills/internlm/wildclawbench/canary"><img src="https://agentmods.dev/badge/skills/internlm/wildclawbench/canary.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00050 | $0.11978 |
| Opus 5 | $0.00025 | $0.05989 |
| Sonnet 5 | $0.00010 | $0.02396 |
| Haiku 4.5 | $0.00005 | $0.01198 |
Grade D, and why
canary scanned grade D with 2 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Strips warnings and disclaimersmediumAnti-refusal
Omitting safety caveats hides risk from the user and is a common jailbreak preamble.
- **Don't lecture.** Offer the fix, explain the risk once, and let the user decide. Avoid repeating warnings or adding disclaimers the user didn't ask for. Reaches for credential fileshighPrivilege escalation
SSH keys, cloud credentials, git-credentials, .npmrc, /etc/shadow: reading these is how a config file becomes a credential leak.
- **Symlink detection**: Check if any files in scanned directories are symlinks pointing to credential files elsewhere on the system. A symlink to `~/.aws/credentials` in a shared workspace is an exposure vector. How it starts
The opening of the file, as written. The whole thing — 719 lines — stays where its author put it; the contents beside it link to each section on GitHub.
🦠Canary
Your agent's early warning system for exposed secrets.
Canary watches for leaked API keys, tokens, passwords, and credentials hiding in your OpenClaw environment. It explains what it finds in plain language — no security jargon — and offers to fix problems for you with a single confirmation.
How It Works
Canary operates in two modes:
🔅 Light Scan (runs automatically on startup)
Every time OpenClaw starts, Canary performs a quick, silent check of the most critical locations:
~/.openclaw/.envand~/.clawdbot/.envfor plaintext credentials- File permissions on config files containing secrets (world-readable = bad)
- Any
.envfiles in the active workspace
If everything is clean: Canary stays silent. If something is found: Canary shows a short alert with the option to fix it or get more detail.
🔠Deep Scan (runs when you ask)
Ask for a full security check whenever you want. The deep scan covers everything in the light scan plus:
- All installed skill directories for hardcoded secrets
- Session/chat history files for accidentally pasted credentials
- Git repositories in the workspace for committed secrets
- SSH keys and config (
~/.ssh/) for weak permissions - Shell history files for commands containing tokens or passwords
- Known credential file paths (
.netrc,.npmrc,.pypirc, Docker config, AWS credentials, etc.)
What Canary Looks For
Canary uses pattern matching and heuristic checks to detect:
| Secret Type | Examples | Where It Looks |
|---|---|---|
| API Keys | Shodan, VirusTotal, OpenAI, Anthropic, AWS, GCP, Stripe, GitHub tokens | .env files, skill configs, shell history, git repos |
| Passwords | Plaintext passwords in configs, database connection strings with embedded passwords | Config files, .env, .netrc, skill directories |
| Private Keys | SSH private keys, PEM files, JWTs with embedded secrets | ~/.ssh/, workspace, skill directories |
| Cloud Credentials | AWS access keys, GCP service account JSON, Azure tokens | ~/.aws/, ~/.config/gcloud/, env vars, configs |
| Tokens & Sessions | OAuth tokens, bearer tokens, session cookies, webhook URLs | Chat history, shell history, .env files |
| Local System Files | Credential exports, service account JSONs, PEM/key files, password manager CSV exports, Kubernetes tokens, Terraform state secrets, database passwords | ~/Downloads/, ~/Desktop/, ~/Documents/, ~/.kube/config, *.tfstate, ~/.config/, ~/Library/Application Support/, ~/.my.cnf, ~/.pgpass, browser password export CSVs, Redis/MongoDB configs |
What ships with it
4 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 719 lines · 50 tokens per session scan D f9d0ea3b18be
canary is a skill published in the GitHub repository InternLM/WildClawBench (516 stars, last pushed 18d ago), licensed MIT. It adds 50 tokens to every session and 11,978 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it D with 2 findings (strips warnings and disclaimers, reaches for credential files). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
security-hardening
Security audit and hardening for AI agents — credential hygiene, secret scanning, prompt injection defense, data leakage prevention, and privacy zones.
007
Security audit, hardening, threat modeling (STRIDE/PASTA), Red/Blue Team, OWASP checks, code review, incident response, and infrastructure security for any project.
infisical-sync-skill
Expert knowledge for the Infisical Sync Hand — Infisical API reference, vault operations, error patterns, security guidance.
contribute-spec-fix
Fix a broken OpenAPI spec in jentic-public-apis with an OpenAPI Overlay, validate it (spectral lint + idempotency check), and contribute it back via a PR to the community catalog. Falls back to applying the same overlay to the local Jentic registry if the user can't wait for maintainer approval, and closes the loop by…
jentic
Use this skill whenever the user wants to work with a third-party or external API/tool through the Jentic platform — e.g. asks to "find the vessel-tracking API and add it", "get rows from this Google Sheet", connect Slack, import/search/discover an API, integrate or automate a SaaS, pull data from a service, or call…
agentsecrets
Zero-knowledge secrets infrastructure — AI agents manage the complete credential lifecycle without ever seeing values.