write-spec

Pass

Audited by Gen Agent Trust Hub on May 9, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches a language reference file from the vendor's GitHub repository (github.com/subsetpark/pantagruel) using WebFetch. This is a legitimate use of external documentation to support the specification design process.\n- [COMMAND_EXECUTION]: Instructions are provided to run the pant CLI tool for type checking and model checking of generated specification files. This is a standard and safe development workflow within the context of formal verification.\n- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface by converting user-provided system descriptions into formal specification code.\n
  • Ingestion points: User responses during the interactive phases (Domain, State, Invariants, etc.) defined in SKILL.md.\n
  • Boundary markers: None implemented in the prompt templates.\n
  • Capability inventory: Generation of .pant files and execution of the pant CLI for verification.\n
  • Sanitization: Not explicitly present; the agent is tasked with translating user intent into formal syntax, which involves inherent interpretation of untrusted input.
Audit Metadata
Risk Level
SAFE
Analyzed
May 9, 2026, 07:33 PM