Back to skill

Security audit

Agent Browser Clawdbot.Disabled

Security checks for vulnerabilities and agentic risk

Overview

This skill is a straightforward browser-automation helper, with disclosed but sensitive auth-state and cookie features that users should handle carefully.

Install only from the intended npm package/source, prefer a pinned reviewed version where possible, and treat auth state files, cookies, and storage output like passwords: do not commit them, share them, or print them unnecessarily, and delete them when no longer needed.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (1)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:194
Finding

Unpinned Global Dependency Installation and Unverified Browser Downloads

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:194-196; duplicated in skills/agent-browser-clawdbot/SKILL.md:194-196
Vulnerability Type: T08: Insecure Dependencies
Risk Level: Medium

Complete Code Snippet:

bash
npm install -g agent-browser
agent-browser install                     # Download Chromium
agent-browser install --with-deps         # Linux: + system deps

Technical Analysis

The installation instructions globally install agent-browser without specifying an exact version or verifying package integrity. Consequently, the code installed depends on whichever package version the npm registry resolves when the command is executed.

The subsequent installation commands download Chromium and may install Linux system dependencies. The project provides no lockfile, checksum, signature-verification procedure, expected registry configuration, or other mechanism for authenticating these downloaded components. Although the metadata identifies the intended upstream repository, that reference does not guarantee that the package and browser artifacts retrieved during installation correspond to an audited release.

This creates a supply-chain trust boundary in which mutable third-party artifacts can execute installation logic and introduce additional components after the skill itself has been reviewed.

Attack Path

  1. An attacker compromises the resolved npm package, its publisher account, the registry distribution path, or a subsequently downloaded browser artifact.
  2. The attacker publishes or substitutes a malicious version while retaining the expected package name.
  3. A user follows the documented npm install -g agent-browser instruction without an exact version or integrity constraint.
  4. npm installs the attacker-controlled package into the user's global environment and may execute package lifecycle scripts with the invoking user's privileges.
  5. The user runs agent-browser install ...[truncated 1272 chars]
Remediation
View remediation

Remediation Suggestions

  1. Pin agent-browser to a specific, reviewed version rather than resolving the latest mutable release:
    bash
    npm install -g agent-browser@EXACT_AUDITED_VERSION
    
  2. Publish and verify the expected npm integrity hash or signed provenance for the selected release.
  3. Document the approved npm registry and reject unexpected registry overrides or similarly named packages.
  4. Prefer a project-local installation with a committed lockfile over a global installation, where operationally feasible.
  5. Pin the Chromium revision and publish authenticated checksums or signatures for downloaded browser artifacts.
  6. Separate browser installation from operating-system dependency installation. Require explicit administrative approval before using --with-deps.
  7. Perform installation in a sandbox, container, or otherwise isolated environment with minimal filesystem and credential access.
  8. Disable or tightly control npm lifecycle scripts when compatible with the package installation process.
  9. Keep the duplicate installation guidance in both SKILL.md files synchronized so that all copies receive the same hardening changes.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (3)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill explicitly exposes commands for reading cookies and storage and for saving/loading authenticated browser state, but it provides no warning that these artifacts may contain session tokens, CSRF secrets, or other credentials. In an agent-facing skill, this omission is dangerous because an agent may persist, inspect, or reuse sensitive session material without user awareness, increasing the risk of credential theft, cross-account access, or accidental leakage into logs and files.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill explicitly instructs users to save and load browser auth state files, which commonly contain cookies, local/session storage, and other session artifacts that can enable account takeover if copied or reused. In an agent context, documenting this workflow without any warning about secure storage, scope, lifetime, or sharing increases the chance that sensitive session material will be persisted insecurely or exfiltrated via logs, artifacts, or shared workspaces.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill documents commands for reading and modifying cookies and browser storage without any caution about the sensitivity of these values. Cookies, tokens, and localStorage/sessionStorage entries often contain authentication or personal data, so exposing these capabilities in agent-facing instructions without guardrails can facilitate credential theft, privacy violations, or unsafe mutation of application state.

Content

No source excerpt is available for this finding.