openclaw-test-performance

Pass

Audited by Gen Agent Trust Hub on Jul 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local shell commands such as pnpm test, /usr/bin/time, and a custom scripts/committer script to perform benchmarks and commit performance improvements. It also utilizes blacksmith testbox for running tests in a remote environment, which is a common practice in CI/CD and distributed testing workflows.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it is instructed to ingest local AGENTS.md files and test result data to guide its optimization decisions. While no boundary markers or sanitization are explicitly defined for this untrusted input, the behavior is standard for development-focused skills and is considered low risk.
  • Ingestion points: Project AGENTS.md files and test execution output (JSON/text).
  • Boundary markers: None present in the instructions.
  • Capability inventory: File reading, shell command execution (pnpm, scripts/committer, blacksmith), and local file writing for reporting.
  • Sanitization: None specified.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 13, 2026, 02:11 PM
Security Audit — agent-trust-hub — openclaw-test-performance