Back to skill

Security audit

Browser Agent Pro

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent browser-automation helper, but it asks users to install a mutable global CLI and load a shared environment file in ways that deserve review before use.

Install only if you are comfortable with a global npm CLI that can control browser sessions. Prefer a pinned/local `agent-browser` install, keep Browserbase credentials in a dedicated file containing only that key, avoid reusing real Chrome profiles unless necessary, and close or clean up saved sessions when finished.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (3)

T01 · Skill Instruction Hijacking

Note
Location
SKILL.md:53
Finding

Unsolicited Promotional Instruction Alters Agent Output

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Error
Location
SKILL.md:19
Finding

Unpinned Third-Party Package Is Installed Globally

Content
View full analysis
Remediation
View remediation
`. - Record and verify package integrity using a lockfile or an equivalent cryptographic integrity mechanism. - Verify the package publisher, source repository, release provenance, and signatures where available. - Prefer a project-local installation over `npm install -g`. - Review package lifecycle scripts and consider disabling them during acquisition when compatible with the package. - Separate browser installation from system dependency installation. - Require explicit user confirmation before installing operating-system packages or making system-wide changes. - Document the exact expected package version consistently in both the metadata and setup instructions. ]]>

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:41
Finding

Shared Environment File Exports Unrelated Secrets to Browser Tooling

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (9)

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

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

  1. User bitten sich anzumelden: https://www.browserbase.com/sign-up (kostenlos)
  2. API Key sicher speichern — User trägt den Key selbst ein:
    bash
    nano ~/.openclaw/.env
    # Zeile hinzufügen: BROWSERBASE_API_KEY=bb_live_XXXXX
    chmod 600 ~/.openclaw/.env
    

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · SKILL.md (reported line 43)May include surrounding context.

  1. User bitten sich anzumelden: https://www.browserbase.com/sign-up (kostenlos)
  2. API Key sicher speichern — User trägt den Key selbst ein:
    bash
    nano ~/.openclaw/.env
    # Zeile hinzufügen: BROWSERBASE_API_KEY=bb_live_XXXXX
    chmod 600 ~/.openclaw/.env
    

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill body is written entirely in German, including operational instructions and warnings, while there is no indication that the skill is intentionally region-specific or that users may choose another language. This can violate a language/locale policy when skills are expected to avoid forcing a specific language without user opt-in.

Content

No source excerpt is available for this finding.

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
80% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · SKILL.md (reported line 43)May include surrounding context.

bash
   nano ~/.openclaw/.env
   # Zeile hinzufügen: BROWSERBASE_API_KEY=bb_live_XXXXX
   chmod 600 ~/.openclaw/.env
  1. Testen:
    bash

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
80% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · SKILL.md (reported line 180)May include surrounding context.

md
| `auth save/login` | Credentials im Auth-Vault gespeichert | Vault-Einträge prüfen (`auth list`), ungenutzte löschen |
| `eval` | Beliebiges JavaScript auf der Seite | Nur auf vertrauenswürdigen Seiten, keine User-Inputs unescaped |
| `clipboard` | Lesen/Schreiben der Zwischenablage | Nur bei Bedarf, Inhalt danach nicht loggen |
| `BROWSERBASE_API_KEY` | Cloud-Zugriff | In `~/.openclaw/.env` mit `chmod 600`, niemals in Logs/Chat wiederholen |

**Generell:** Bevorzuge isolierte Sessions (`--session <name>`) statt echte Chrome-Profile. Schließe Browser nach Gebrauch (`close --all`). Speichere keine Secrets in Skill-Dateien.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

This reference documents authentication/session commands such as loading saved state, reusing Chrome profiles, and saving/restoring auth profiles without any warning about the sensitivity of cookies, tokens, and persisted browser storage. In an agent skill context, these commands can enable credential reuse, unintended cross-session access, or leakage of authenticated state if an operator follows the reference naively.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This section lists commands that expose or capture sensitive browser data, including reading cookies/local storage, screenshots/PDF capture, and especially retrieving a CDP WebSocket URL, but provides no warning about confidentiality or takeover risk. In browser automation, CDP access can grant deep control of an existing browser session, and screenshots or storage inspection can capture secrets, PII, and active-session artifacts.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
93% confidence
Finding

The command reference is written entirely in German, starting with the introductory sentence and continuing through all command descriptions, without offering an alternative language or documenting that the skill is intentionally region-specific. This may violate a language/locale policy requiring user choice or explicit justification.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
90% confidence
Finding

Clipboard read/write commands are documented without warning even though the clipboard often contains passwords, tokens, personal data, or proprietary content. In an agent setting, silent clipboard access can lead to unintended disclosure or tampering with user data across application boundaries.

Content

No source excerpt is available for this finding.