edit-attractor-pipelines
Pass
Audited by Gen Agent Trust Hub on Jun 28, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes DOT files via a proof command to verify workflow policy, which involves running code or configurations from the target repository.
- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface. Ingestion points: Reads AGENTS.md, attractor-spec.md, and DOT files from repositories. Boundary markers: Absent; there are no instructions to use delimiters for external data. Capability inventory: Can execute proof commands, commit to git, and create pull requests. Sanitization: Absent; the skill does not specify any validation for ingested data.
Audit Metadata