verify-this
Pass
Audited by Gen Agent Trust Hub on Aug 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute code behavior verification through focused tests or minimal reproduction scripts. It also utilizes CLI tools (e.g.,
control-cli) and UI controllers (control-ui) to capture evidence, which involves running commands in the agent's environment. - [DATA_EXPOSURE]: The workflow involves capturing and storing artifacts such as heap snapshots, HTTP responses, and screenshots which may contain sensitive information. The skill includes a specific guideline to keep evidence inline and minimize disk storage in
/tmp/verify-this/unless the user explicitly agrees, demonstrating awareness of data sensitivity. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data from external sources, including terminal transcripts, HTTP responses, and test outputs. This creates a surface where malicious instructions embedded in the processed data could potentially influence agent behavior.
- Ingestion points: Terminal transcripts, HTTP responses, heap snapshots, and test outputs are read into the agent context.
- Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions for the processed artifacts.
- Capability inventory: The skill uses subprocess execution via CLI tools and file system write operations for artifact storage.
- Sanitization: There are no explicit instructions for filtering or sanitizing the content of the captured artifacts before comparison.
Audit Metadata