create-formal-verification-skill

Pass

Audited by Gen Agent Trust Hub on Oct 29, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands for installing and running verification tools. This is a primary function of the skill for performing formal checks.
  • [EXTERNAL_DOWNLOADS]: The instructions guide the agent to utilize external model checkers and verifiers such as TLA+. These are well-known and trusted tools within the software verification community.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted codebase content to define verification properties, creating an attack surface for instructions hidden in source code.
  • Ingestion points: Repository inspection for property and logic analysis (SKILL.md).
  • Boundary markers: No explicit delimiters or ignore-instructions warnings are defined for source code processing.
  • Capability inventory: Execution of shell commands and creation of new skill files (SKILL.md).
  • Sanitization: Relies on the agent's translation of code to formal logic models.
  • [DYNAMIC_EXECUTION]: The skill generates project-local skill files and verification scripts based on the analysis performed. This is a low-risk use of dynamic instruction generation intended for task persistence.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 29, 2026, 12:20 PM