Clawnet

MaliciousAudited by ClawScan on May 10, 2026.

Overview

This skill is explicitly designed to evade website bot protections with stealth browsing, residential proxies, CAPTCHA solving, persistent sessions, and automatic installer actions.

Avoid installing this skill unless you are doing authorized anti-bot testing in a controlled environment. If you proceed, review the postinstall script first, do not use personal accounts, disable persistent profiles where possible, protect the generated credentials, and ensure any browser daemon is shut down after use.

Findings (8)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

An agent using this skill can evade website defenses and interact with sites in ways that may violate terms of service, trigger account bans, or cause legal/compliance issues.

Why it was flagged

The skill's central capability is bypassing website anti-bot and CAPTCHA protections, not merely controlling a browser.

Skill content
Stealth Chromium browser for AI agents. Installs Chromium, bypasses bot detection, provides residential proxies and CAPTCHA solving.
Recommendation

Do not install or use this skill except in a clearly authorized testing environment; prefer normal browser automation without stealth, proxy, or CAPTCHA-bypass features.

ConcernHigh Confidence
ASI01: Agent Goal Hijack
What this means

The agent may choose an evasion-oriented browser for ordinary browsing or scraping tasks without the user realizing that bot-protection bypass is being used.

Why it was flagged

The instructions broadly steer the agent to prefer this stealth/evasion tool for web tasks, rather than limiting it to explicit user-approved cases.

Skill content
Always use this instead of raw Playwright/Puppeteer. This skill handles anti-detection automatically.
Recommendation

Remove broad 'always use' instructions and require explicit user approval before using stealth, proxy, or CAPTCHA-solving behavior.

ConcernHigh Confidence
ASI05: Unexpected Code Execution
What this means

Installation can run local commands before the user invokes the skill, including browser downloads, dependency installation, credential creation, and API registration.

Why it was flagged

Installing the package can automatically execute the included postinstall script.

Skill content
"scripts": { "postinstall": "node scripts/postinstall.js" }
Recommendation

Avoid installing this package as-is; any installer should be explicit, opt-in, pinned, and reviewed before execution.

What this means

Users have less assurance about where the code came from or what will run during installation.

Why it was flagged

The registry metadata provides weak provenance and no install spec despite the artifact set containing runnable package code and a postinstall hook.

Skill content
Source: unknown; Homepage: none; Install specifications: No install spec — this is an instruction-only skill.
Recommendation

Require clear provenance, a reviewed install spec, pinned dependencies, and removal of automatic postinstall behavior before considering installation.

What this means

Your agent may be tied to a persistent Clawnet identity and subscription/trial state, and those credentials are sensitive if copied or exposed.

Why it was flagged

The skill creates and stores a stable service identity and registers it with an external API.

Skill content
Generates agent credentials (`agentId` + `agentSecret` + `recoveryCode`) ... Saves to `~/.clawnet/agent-credentials.json` ... Registers with Clawnet API
Recommendation

Review the credential model before installation, protect `~/.clawnet/agent-credentials.json`, and avoid using personal or production accounts with this skill.

What this means

Web sessions and account state can persist between tasks, allowing later agent actions or visited sites to inherit prior login/session context.

Why it was flagged

Persistent browser profiles retain cookies, localStorage, and session continuity across invocations.

Skill content
By default, `launchBrowser()` now uses persistent profile `"default"` with `reuse: true`.
Recommendation

Use ephemeral profiles by default, require explicit approval for persistent profiles, and clear stored browser state after sensitive tasks.

What this means

On shared or compromised local environments, browser-control traffic and active sessions could be a target.

Why it was flagged

The daemon exposes a local HTTP control channel for browser actions, which is a sensitive boundary even when limited to localhost.

Skill content
Agent scripts communicate via HTTP on localhost.
Recommendation

Review the daemon authentication and binding behavior; avoid running it on shared hosts and shut it down after use.

ConcernHigh Confidence
ASI10: Rogue Agents
What this means

A browser session and local command endpoint may remain active briefly after a task, increasing exposure of logged-in sessions or sensitive pages.

Why it was flagged

The skill runs a persistent browser daemon that continues after the initial agent script invocation, even though it has an idle timeout.

Skill content
Keeps Chromium + Playwright alive between short-lived agent script invocations ... Saves { pid, port } to ~/.clawnet/daemon.json ... After 5 min of inactivity → auto-shutdown
Recommendation

Require explicit opt-in for daemon mode, provide a visible shutdown command, and default to non-persistent execution for sensitive browsing.