chromium-browser

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to use curl for interacting with the Wayback Machine CDX and Availability APIs. It also performs local network requests to the Chromium remote-debugging port (default 9222) to verify connectivity.
  • [EXTERNAL_DOWNLOADS]: The skill fetches capture metadata and archival content from the Internet Archive's Wayback Machine (archive.org and web.archive.org), which are well-known services. These downloads are restricted to data retrieval and research purposes.
  • [DATA_EXFILTRATION]: Network activity is limited to authorized Wayback Machine API queries and local Chromium debugging. The skill explicitly forbids bypassing network-header redaction or seeking credentials through process introspection.
  • [PROMPT_INJECTION]: While the skill ingests external content from websites and APIs, it provides clear boundaries to mitigate indirect prompt injection risks.
  • Ingestion points: Browser snapshots, screenshots, and JSON responses from Wayback Machine APIs.
  • Boundary markers: Mandatory use of specific pageId handles for every scoped tool and instructions to avoid surfacing unrelated browser state.
  • Capability inventory: Includes browser navigation, JavaScript evaluation (evaluate_script), and snapshot generation.
  • Sanitization: The skill relies on structured snapshots and manual verification rather than raw data execution, and it restricts file output to authorized paths.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 11:59 AM
Security Audit — agent-trust-hub — chromium-browser