iterative-loop

Pass

Audited by Gen Agent Trust Hub on Sep 18, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill utilizes the 'ruflo' package via 'npx --no-install'. While the flag prevents new installations if not present, it still introduces a dependency on a non-standard external tool to manage hooks and memory states.
  • [COMMAND_EXECUTION]: The skill's primary functionality involves executing shell commands for verification and orchestration, including 'npm test', 'eslint', 'tsc', and 'npx'. These are used within autonomous loops to determine completion.
  • [INDIRECT_PROMPT_INJECTION]: The skill implements a self-correcting development cycle where the agent's next steps are guided by the output of external tools and previously written code.
  • Ingestion points: Test results from shell commands, git history, and file contents processed during iterations (SKILL.md).
  • Boundary markers: The skill suggests using explicit success criteria and 'promise' tags as boundaries, which help but do not fully mitigate adversarial input in processed files.
  • Capability inventory: Shell execution capabilities for testing, linting, and state management ('npm', 'npx', 'tsc', 'eslint').
  • Sanitization: The instructions do not specify sanitization or escaping of tool outputs before they are reintroduced into the prompt for the next iteration step.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 18, 2026, 06:15 PM
Security Audit — agent-trust-hub — iterative-loop