verification-loop

Pass

Audited by Gen Agent Trust Hub on Oct 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses shell commands to run project build and verification scripts such as npm run build, pnpm build, npm run test, and npm run lint. It also utilizes standard utilities like npx tsc, pyright, and ruff for code analysis. These operations are appropriate for a developer-oriented agent.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external tool outputs and file contents into the agent's context, creating a potential surface for indirect prompt injection.\n
  • Ingestion points: The agent reads outputs from various verification tools and file differences via git diff and grep as described in SKILL.md.\n
  • Boundary markers: There are no specific delimiters or instruction wrappers used to segregate untrusted data from the agent's core instructions.\n
  • Capability inventory: The agent environment has capabilities for shell command execution and file system manipulation.\n
  • Sanitization: The skill does not implement sanitization or filtering of the ingested data before it is added to the model's context window.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 14, 2026, 02:58 PM