setup-harness

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The setup instructions include a template for AGENTS.md that contains a command to execute a local PowerShell script (pwsh -File path/to/agent-harness/install.ps1) for environment synchronization.
  • [INDIRECT_PROMPT_INJECTION]: The skill manages the initialization of project-level instruction files (AGENTS.md, CLAUDE.md).
  • Ingestion points: User input for project configuration (tracker type, RAG status) in SKILL.md.
  • Boundary markers: Absent; harness instructions are appended as new sections.
  • Capability inventory: File writing to repository documentation paths and local script execution.
  • Sanitization: Absent; configuration values are interpolated directly into the provided templates.
  • [EXTERNAL_DOWNLOADS]: The documentation contains a reference link to a GitHub repository (github.com/mattpocock/skills) which served as the basis for this harness setup.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 11:52 PM
Security Audit — agent-trust-hub — setup-harness