Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

Agent Browser Xyhh

v1.0.0

A fast Rust-based headless browser automation CLI with Node.js fallback that enables AI agents to navigate, click, type, and snapshot pages via structured co...

0· 50·1 current·1 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
The name/description (headless browser CLI) aligns with the runtime instructions (lots of agent-browser CLI commands). Requiring node/npm is reasonable because the SKILL.md recommends installing an npm package. However there are mismatches: the description mentions a Rust-based CLI with Node fallback, while the instructions and installation recommend an npm package; and SKILL.md's 'From Source' steps reference git and pnpm even though those binaries are not listed in the declared required binaries. Also _meta.json ownerId/version differ from the registry metadata, which suggests sloppy or inconsistent packaging.
Instruction Scope
The SKILL.md tells the agent to run an external CLI (agent-browser) to navigate pages, take screenshots, record video, manipulate cookies/storage, and upload files — all expected for a browser-automation tool. The instructions do not explicitly tell the agent to read arbitrary local files or environment variables beyond the CLI usage, but they do recommend installing packages from npm and cloning from GitHub. The instructions reference commands (git, pnpm) that aren't declared as required binaries.
Install Mechanism
There is no formal install spec in the skill bundle (instruction-only). The SKILL.md recommends installing agent-browser via npm -g or building from source via git/pnpm. Installing a package from the public npm registry is a common pattern, but it does involve fetching and running third-party code from the network (moderate risk). No suspicious download URLs are present in the docs, but the skill bundle does not declare or pin a verified source or checksum.
Credentials
The skill does not request environment variables, credentials, or config paths. The functionality described (web navigation, screenshots, cookies/storage) typically does not require additional host secrets. This is proportionate to the stated purpose.
Persistence & Privilege
always:false and disable-model-invocation:false (normal). The skill does not request persistent system-wide configuration or claim it will modify other skills. No high-privilege flags are requested.
What to consider before installing
This skill appears to be a wrapper around an external 'agent-browser' CLI and mostly does what it says, but there are a few inconsistencies and network-install risks to consider before installing: - Metadata mismatches: the included _meta.json ownerId/version do not match the registry metadata; this could be a packaging mistake or indicate the bundle wasn't maintained carefully. Ask the publisher or check the skill registry entry for provenance. - Missing declared tools: SKILL.md suggests using git and pnpm (and running npm -g), but the skill's declared required binaries only list node and npm. If you plan to build from source, ensure git and pnpm are available and trusted. - Network install: the instructions recommend npm install -g agent-browser which will download code from the npm registry. Verify the package name, author, and npm page. Prefer installing in a controlled environment (container, VM) rather than on a critical machine. - Review upstream: the README points to github.com/vercel-labs/agent-browser — inspect that repository, confirm maintainers, and check package contents and release signatures if possible. - Data-exfiltration risk: any browser automation tool can be used to visit pages and extract data. Be cautious when running it against sensitive sites or when allowing the agent to navigate pages containing secrets. If you want to proceed: validate the npm package and repository, avoid installing globally on a production host, and run the tool in an isolated environment until you are confident in its provenance.

Like a lobster shell, security has layers — review code before you run it.

latestvk978we5m8xd3qet58ngcwddm7s83jr91

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

Runtime requirements

🌐 Clawdis
Binsnode, npm

Comments