iterate-agent-harness
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to create executable scripts (e.g.,
scripts/test-completion-gate.mjs) and run them or the project's test suite (e.g.,npm test). While expected for its purpose, this allows the agent to execute shell commands and custom logic. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and interpret data from potentially untrusted sources, including agent transcripts, CI logs, and state files. This creates an attack surface where malicious content within a log file could attempt to influence the agent's behavior during the interpretation and classification phases.
- Ingestion points: Agent transcripts,
.cursor/*-state.mdfiles,git diffoutput, and CI logs. - Boundary markers: The instructions do not define explicit delimiters or 'ignore' warnings for the external data being processed.
- Capability inventory: The agent has the ability to write to the filesystem (
SKILL.md,CHANGELOG.md,scripts/) and execute commands (npm test). - Sanitization: There is no mention of escaping or validating the content of the captured evidence before processing.
- [DYNAMIC_EXECUTION]: Phase 2 and 3 involve the dynamic generation and subsequent execution of new validation logic (hooks/scripts). This is a core feature of the skill but represents a runtime code generation surface.
Audit Metadata