code-overview
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/common.pyusessubprocess.runto callgitfor retrieving commit information. This is performed using a list of arguments and is restricted to retrieving metadata from the local repository being analyzed. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from the repository's configuration files (like
package.jsonorpyproject.toml) and third-party analysis reports. - Ingestion points: Repository manifests and external JSON reports are read in
scripts/discover_packages.pyandscripts/common.py. - Boundary markers: Data from manifests is interpolated into reports; the skill relies on the structured nature of JSON and TOML.
- Capability inventory: Includes file system write access for generating documentation and
subprocess.runfor git metadata retrieval. - Sanitization: The skill consistently uses HTML escaping (
html.escape) and JSON string sanitization to prevent injection into the generated HTML reports.
Audit Metadata