cursor-maintain-formal-verification-skill
Pass
Audited by Gen Agent Trust Hub on Oct 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from git commit messages and code diffs to determine how to update verification models and properties. An attacker could craft malicious commit messages or code comments to influence the agent's logic during the synchronization process.
- Ingestion points: Git commit messages and file diffs between the last sync commit and the current HEAD (SKILL.md).
- Boundary markers: None; instructions do not provide explicit delimiters or warnings to ignore instructions embedded within the processed commit history.
- Capability inventory: The agent is authorized to modify verification files (models, harnesses, properties) and execute shell commands to run verifiers (SKILL.md).
- Sanitization: None; the agent is instructed to read and interpret commit messages to decide if a "promise changed on purpose" or if the "verification no longer describes the code."
- [DYNAMIC_EXECUTION]: The skill dynamically identifies and executes shell commands that are retrieved from project-local skills or configuration files within the repository.
- Evidence: The instructions in SKILL.md direct the agent to "collect... the command that runs it" and "Run it as it is" to verify the code. This results in the execution of arbitrary commands defined in the local environment's configuration files.
Audit Metadata