maintain-formal-verification-skill

Pass

Audited by Gen Agent Trust Hub on Oct 26, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow that requires ingesting untrusted external data and using it to perform actions, which creates a potential injection surface.
  • Ingestion points: The skill reads user-defined targets, covered-code changes, property files, and verification instructions from the project environment (referenced in Step 1 and Step 2 of SKILL.md).
  • Boundary markers: Absent. There are no explicit instructions or delimiters used to ensure the agent ignores potentially malicious commands or instructions embedded within the code or verification files it analyzes.
  • Capability inventory: The skill instructs the agent to execute shell commands to run external verification tools (model checkers, provers, verifiers) and perform code mutations in scratch worktrees (Step 1, Step 2, and Step 4 of SKILL.md).
  • Sanitization: Absent. The instructions do not specify a validation or sanitization process for the file content or the commands before they are executed.
  • [COMMAND_EXECUTION]: The skill requires the execution of external tools and shell commands as part of its primary function.
  • Evidence: Step 1 and Step 4 in SKILL.md direct the agent to find and run model checkers, provers, and verifiers, as well as manage isolated scratch copies or worktrees for fault injection testing.
  • [SAFE]: The skill contains references to an external upstream source used for maintenance and tracking.
  • Evidence: UPSTREAM.md identifies the canonical source as the Cursor plugins repository on GitHub, which is a well-known service. These references are documented neutrally and follow standard supply chain tracking practices.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 26, 2026, 01:27 PM