evals
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill clones a private test repository from the author's GitHub account (johnviklund/j-skills-evals) to the local filesystem (~/.agents/j-skills-evals).
- [COMMAND_EXECUTION]: The skill instructs the agent to use gh and git CLI tools to manage the test repository and commit evaluation results.
- [INDIRECT_PROMPT_INJECTION]: The skill processes markdown files containing diffs and context from the external test repository.
- Ingestion points: Markdown files in ~/.agents/j-skills-evals/strict-reviewer/.
- Boundary markers: The instructions direct the agent to provide specific sections (Context, Diff) to the model being evaluated, but do not specify explicit delimiters.
- Capability inventory: Capabilities include network access via gh/git and file writing to RESULTS.md.
- Sanitization: No explicit sanitization of the test case content is described.
Audit Metadata