python-authoring
Pass
Audited by Gen Agent Trust Hub on Jul 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill establishes a development standard for Python 3.12, prioritizing the standard library and repository-local packaging conventions.
- [PROMPT_INJECTION]: No instructions were found that attempt to bypass safety guardrails or override agent behavior.
- [DATA_EXFILTRATION]: The instructions explicitly forbid network access and external state dependencies during testing and validation phases.
- [COMMAND_EXECUTION]: The skill references the use of local build and check tools (
just bundle,just check) which are standard within the described development environment. - [INDIRECT_PROMPT_INJECTION]: The skill mandates a secure data handling pattern by requiring that all untrusted input be validated at the boundary and converted into typed, trusted data structures (dataclasses) before processing.
- Ingestion points: The skill instructs the agent to read repository configurations, callers, and existing conventions (SKILL.md).
- Boundary markers: The 'Model and validate data explicitly' section requires explicit validation at boundaries.
- Capability inventory: The skill uses local shell commands for building and testing code via a task runner.
- Sanitization: The skill enforces the use of standard library parsers and type-checking for all external data.
Audit Metadata