The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local git commands using subprocess.Popen and subprocess.run to extract commit history and metadata. These calls use argument lists rather than shell strings, which is a secure practice that prevents shell injection. This functionality is core to the skill's purpose of analyzing git repositories.
Evidence found in scripts/build_ownership_map.py, scripts/community_maintainers.py, and scripts/run_ownership_map.py.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from git logs, including author names, email addresses, and file paths. This metadata is summarized for the agent, creating an attack surface where adversarial strings in the git history could attempt to influence the agent's behavior when reviewing the summary.
Ingestion points: Git history data is ingested and parsed in scripts/build_ownership_map.py and scripts/community_maintainers.py.
Boundary markers: The skill does not employ explicit boundary markers or instructions to ignore embedded commands within the processed git data.
Capability inventory: The skill's execution environment includes access to Bash, Write, and Edit tools as defined in the frontmatter configuration.
Sanitization: While the skill parses data into structured CSV and JSON, it does not explicitly filter or sanitize the content for potential prompt injection patterns.

openai-security-ownership-map