cultivar
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes the
cultivarCLI to perform tasks such asinit,run,grade, andreport, which are standard operations for its stated purpose of agent evaluation. - [COMMAND_EXECUTION]: Task configuration files (
tasks/*.yaml) allow for optional shell hooks (setup,teardown,verify) to be executed during the test lifecycle. - [EXTERNAL_DOWNLOADS]: The tool can fetch and cache documentation from external
http(s)://URLs specified in thecontext_refsorself_navigate_refsfields of a task. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data that could contain malicious instructions.
- Ingestion points: URLs in
context_refsandself_navigate_refs, andintentfields intasks/*.yaml. - Boundary markers: None explicitly mentioned in the skill documentation; users are advised to use
--dry-runfor inspection. - Capability inventory: Execution of shell commands via
cultivarand network requests to LLM providers. - Sanitization: None described in the instructions beyond manual review using
--dry-run.
Audit Metadata