astcount-refactor-loop
Pass
Audited by Gen Agent Trust Hub on Aug 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted code and executes shell commands to run tests, creating a surface for injection.
- Ingestion points: The agent reads and refactors source code within the repository specified in the prompt.
- Boundary markers: No delimiters are used to wrap the ingested code to prevent the agent from following embedded instructions.
- Capability inventory: The skill executes shell commands via nix, bunx, or npx to run astcount, and it executes arbitrary test commands as discovered in the repository or provided by the user.
- Sanitization: There is no evidence of sanitization or validation of the code contents or the test execution commands before they are run.
- [EXTERNAL_DOWNLOADS]: The skill fetches the
astcountutility from the vendor's GitHub repository or official package registries during execution. - [COMMAND_EXECUTION]: Shell commands are used to invoke complexity analysis tools and to verify the behavior of refactored code through automated testing.
Audit Metadata