python-canary-fix
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [SAFE]: The skill is a dedicated utility for repository maintenance, authored by 'arize-ai' for their own 'openinference' project. Its operations are properly scoped to this environment.
- [COMMAND_EXECUTION]: The skill executes
toxtests usinguvx. This is standard professional practice for Python package development and testing across multiple environments, posing no unexpected security risk in this context. - [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The skill reads failure logs from GitHub Actions via the
gh run view --log-failedcommand inSKILL.md. - Boundary markers: No explicit delimiters or safety instructions are used when processing the log content.
- Capability inventory: The skill has the ability to modify source code, execute local tests using
tox, and create pull requests on GitHub. - Sanitization: No specific sanitization of the log content is performed before analysis.
- Analysis: Although the skill processes external data from CI logs, the risk is mitigated by its specific focus on a trusted repository environment and its purpose of assisting developer maintenance tasks.
Audit Metadata