dependency-doctor

Pass

Audited by Gen Agent Trust Hub on Nov 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill can fetch package metadata from the official PyPI registry (pypi.org) when the --online flag is used. This is a well-known service and the data is used exclusively to identify yanked releases.
  • [COMMAND_EXECUTION]: The skill executes a local Python script (scripts/dep_doctor.py) using the python3 command to perform manifest analysis. This script uses only the Python standard library and does not execute arbitrary code.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes external data from dependency manifest files which could potentially contain malicious instructions.
  • Ingestion points: Reads contents of requirements.txt, pyproject.toml, and package.json from the local file system.
  • Boundary markers: The script outputs a structured JSON report which the agent then interprets, providing a layer of separation between the raw file content and the agent's prompt.
  • Capability inventory: The agent has the capability to read files, execute the included analysis script, and propose file edits (which require explicit user approval).
  • Sanitization: The Python script uses regular expressions to parse package names and versions, and it explicitly strips comments from Python requirements files, reducing the risk of hidden instructions being processed.
  • [SAFE]: No obfuscation, persistence mechanisms, or privilege escalation attempts were detected. The skill follows the principle of least privilege by asking for permission before performing network operations or applying file fixes.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 14, 2026, 09:17 PM