deliver-setup

Pass

Audited by Gen Agent Trust Hub on Nov 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill performs environment diagnostics by executing several CLI tools, including git, gh, herdr, and orca. It also runs a Python script, deliver-runtime.py, which is bundled with the skill to manage host selection and configuration.
  • [EXTERNAL_DOWNLOADS]: The skill guides the user to install dependencies from external, third-party GitHub repositories such as mattpocock/skills, herdrdev/herdr, stablyai/orca, and getpaseo/paseo. These are community-maintained resources and are not recognized as trusted vendors.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes project-specific configuration files (delivery.toml, AGENTS.md) and external markdown references. This creates a surface for indirect prompt injection if those files contain malicious instructions.
  • Ingestion points: docs/agents/delivery.toml, AGENTS.md, and shared configuration references.
  • Boundary markers: The skill does not implement specific boundary markers or instructions to disregard embedded commands in the processed data.
  • Capability inventory: The skill possesses the ability to modify repository files, perform Git commits, and execute local Python scripts (deliver-runtime.py, deliver-orca-preflight.py).
  • Sanitization: No sanitization or strict schema validation is mentioned for the ingested configuration data.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 12, 2026, 05:36 AM