maintain-formal-verification-skill

Pass

Audited by Gen Agent Trust Hub on Nov 8, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze source code, formal specifications, and file-line citations to detect model drift. This represents a potential ingestion surface where malicious content within the analyzed code could attempt to influence agent behavior.
  • Ingestion points: The agent reads source code files, formal specification documents, and line-based anchors from the project directory.
  • Boundary markers: The skill does not provide specific instructions for using delimiters or boundary markers to distinguish between data being analyzed and the agent's operational instructions.
  • Capability inventory: The agent is authorized to execute complex verification tools (TLC, apalache, kani) and write to a ledger via an external JavaScript module.
  • Sanitization: The skill does not specify any sanitization or validation routines for the ingested code before processing.
  • [COMMAND_EXECUTION]: The workflow requires the agent to run various shell-based tools, including 'TLC', 'apalache', 'lake build', and 'cargo kani'. While these are standard verification utilities, their execution is triggered by analysis of the project's state, which is a common pattern for interacting with development environments.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 8, 2026, 07:21 PM