Back to skill

Security audit

Agent Browser Cn

Security checks for vulnerabilities and agentic risk

Overview

This browser automation skill is mostly coherent, but it exposes authenticated browser state, cookies, recordings, traces, and an unpinned global install path without enough safety guidance.

Review carefully before installing. Use it only in a contained environment, avoid running it against sensitive accounts unless necessary, treat saved state files, screenshots, videos, PDFs, traces, cookies, and localStorage as secrets, delete them after use, and prefer a pinned, reviewed local install over the documented global latest install.

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:20
Finding

Unpinned Global npm Installation Creates a Supply-Chain Execution Risk

Content
View full analysis

Vulnerability Details

File Locations: SKILL.md:20-22; CONTRIBUTING.md:23-25
Vulnerability Type: Unpinned and mutable third-party dependency installation
Risk Level: Medium

Vulnerable Code

SKILL.md:20-22:

bash
npm install -g agent-browser
agent-browser install
agent-browser install --with-deps

CONTRIBUTING.md:23-25:

bash
npm install -g agent-browser@latest

Technical Analysis

The installation instructions retrieve and globally install agent-browser without pinning an exact reviewed version. The troubleshooting instructions explicitly select @latest, making the installed artifact dependent on the package registry state at execution time.

npm installation can execute package lifecycle scripts with the privileges of the invoking user. A global installation also places executable files into shared user-level or system-level npm locations. Furthermore, agent-browser install --with-deps may retrieve and install additional mutable components without documented versions, checksums, or integrity-validation procedures.

The audited project does not itself contain a malicious payload, and the evidence does not establish that the current upstream package is malicious. The vulnerability is the absence of controls that ensure users receive the same reviewed dependency release over time.

Attack Path

  1. An attacker compromises the upstream npm package, a maintainer account, or another component in the dependency publication chain.
  2. The attacker publishes a malicious release under the expected package name.
  3. A user or AI agent follows the documented npm install -g agent-browser, npm install -g agent-browser@latest, or dependency-installation instructions.
  4. npm resolves the mutable package reference to the attacker-controlled release.
  5. Malicious lifecycle code executes during installation, or a compromised globally installed executable runs when subsequent d ...[truncated 999 chars]
Remediation
View remediation

Remediation Suggestions

  1. Replace mutable package references with an exact version that has been reviewed:

    bash
    npm install --save-exact agent-browser@<reviewed-version>
    
  2. Remove the recommendation to install @latest. Document a controlled upgrade and security-review process instead.

  3. Prefer a project-local dependency over a global installation, and commit a lockfile so transitive versions are reproducible.

  4. Verify downloaded artifacts using registry integrity metadata, checksums, or signed provenance before execution.

  5. Document the exact components and versions installed by agent-browser install and agent-browser install --with-deps.

  6. Run installation without administrative privileges and in an isolated environment where practical.

  7. Consider disabling npm lifecycle scripts during initial acquisition and reviewing the package before enabling any required installation scripts.

  8. Periodically audit the pinned package and its transitive dependencies, then update pins only after review.

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • Rogue AgentSelf-Modification, Session Persistence
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

Self-Modification

High
Category
Rogue Agent
Confidence
85% confidence
Finding

Skill modifies its own code, configuration, or behavior at runtime. Self-modification enables an agent to escalate privileges, disable safety constraints, or install persistent backdoors.

Content

Scanner excerpt · CONTRIBUTING.md (reported line 60)May include surrounding context.

md
## Adding New Commands to the Skill

Update SKILL.md when the upstream CLI adds new commands.
- Keep the Installation section
- Add new commands in the correct category
- Include usage examples

Context Window Stuffing

Medium
Category
Memory Poisoning
Confidence
85% confidence
Finding

Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.

Content

Scanner excerpt · SKILL.md (reported line 41)May include surrounding context.

agent-browser open # Navigate to page agent-browser snapshot -i # Get interactive elements with refs agent-browser click @e1 # Click element by ref agent-browser fill @e2 "text" # Fill input by ref agent-browser close # Close browser

text

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill documents screenshots, PDFs, video recording, console logs, and traces without noting that these artifacts may capture credentials, personal data, tokens, page content, or debugging secrets. In browser automation workflows, such artifacts are commonly persisted or shared, so omitting handling guidance increases the risk of sensitive data leakage.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill explicitly encourages saving and loading browser session state but does not warn that the state file can contain authentication cookies, bearer tokens, and other sensitive session artifacts. In an agent environment, this can lead to credential theft, unintended privilege reuse across tasks, or accidental disclosure if the file is stored insecurely or reused in the wrong context.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
94% confidence
Finding

Loading saved browser state enables session persistence across runs, which can silently transfer authenticated context between tasks, users, or trust boundaries. Without warnings or safeguards, this increases the chance of unauthorized account access, stale privileged sessions, or replay of sensitive authenticated state.

Content

Scanner excerpt · SKILL.md (reported line 236)May include surrounding context.

bash
agent-browser state save auth.json    # Save session state
agent-browser state load auth.json    # Load saved state

Example: Form submission

Session Persistence

Medium
Category
Rogue Agent
Confidence
94% confidence
Finding

The authentication example normalizes reusing saved state for later sessions without highlighting that the file may preserve active login context. In agent-driven environments, this makes accidental privilege carryover and credential-equivalent file exposure more likely, especially if artifacts are reused broadly.

Content

Scanner excerpt · SKILL.md (reported line 265)May include surrounding context.

agent-browser wait --url "/dashboard" agent-browser state save auth.json

Later sessions: load saved state

agent-browser state load auth.json agent-browser open https://app.example.com/dashboard

text