eval-harness

Pass

Audited by Gen Agent Trust Hub on Oct 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill defines a Model-Based Grader that uses the agent to evaluate task outputs, creating a surface for embedded instructions in the data to influence the agent. 1. Ingestion points: Task descriptions and code outputs ingested for evaluation in the Grader Types section of SKILL.md. 2. Boundary markers: The provided [MODEL GRADER PROMPT] template lacks specific delimiters or instructions to treat input data as untrusted. 3. Capability inventory: The skill utilizes Read, Write, Edit, Bash, Grep, and Glob tools. 4. Sanitization: No sanitization or filtering of the content being evaluated is described in the workflow.
  • [COMMAND_EXECUTION]: The skill utilizes shell commands to perform deterministic checks for task completion. Evidence: Examples in SKILL.md demonstrate the use of the Bash tool to execute npm test, npm run build, and grep to verify project status.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 14, 2026, 02:58 PM