migrating-pi-to-pydantic-ai
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill requires the AI agent to read and analyze untrusted external data, including repository instructions, source code, package manifests, and session history logs. This analysis surface could be exploited if the analyzed content contains instructions designed to manipulate the agent's reasoning during the migration process.
- Ingestion points: Instructions in
SKILL.md(Step 1: "Read repository instructions, package.json, Pi settings, tests, extension/package manifests, and runtime entrypoints.") andreferences/VERIFICATION-AND-CUTOVER.md(characterizing inputs and messages). - Boundary markers: The skill does not prescribe the use of specific boundary markers or delimiters for the agent to use when isolating content from the external files it analyzes.
- Capability inventory: The agent is assumed to have read access to the local file system to process the source repository. The skill also discusses the implementation of
ShellandFileSystemcapabilities in the target application. - Sanitization: No specific methods for sanitizing or escaping content from the external repositories are mentioned for the agent's analysis phase.
Audit Metadata