har-derived-api-client

Pass

Audited by Gen Agent Trust Hub on Sep 27, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The scripts har_capture.py and har_capture_cdp.py use the Playwright library to perform browser automation tasks such as navigation, clicking, and form filling based on user-provided arguments.
  • [EXTERNAL_DOWNLOADS]: The skill requires the playwright package, which facilitates the download of official browser binaries (e.g., Chromium). This is standard functionality for the library and originates from trusted sources.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from web pages and displays it in the terminal, creating a potential attack surface. However, this is consistent with the skill's purpose as an analysis tool.
  • Ingestion points: Network requests and responses are captured from external sites and stored in HAR files, then read by har_to_client.py.
  • Boundary markers: The output uses clear text headers to separate different API endpoints, providing basic visual delimitation.
  • Capability inventory: The skill allows for browser automation and file writes to the local system, but does not automate the execution of derived API calls.
  • Sanitization: Content samples in the analysis output are truncated using the --max-body parameter to prevent excessive data from being processed.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 27, 2026, 04:03 PM