grant-thinking-general
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions include specific shell commands for the agent to execute to check for updates (
git ls-remote,awk,sed,sort,tail) and to perform a self-update (git pull --ff-only). These commands are documented as a maintenance feature and are contingent on explicit user consent. - [EXTERNAL_DOWNLOADS]: The skill fetches version tags and potentially new code from the remote repository defined as 'origin'. This functionality is tied to the vendor's own infrastructure for skill updates.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted project ideas, draft logic, and proposal sections provided by users.
- Ingestion points: User-provided grant ideas, proposal concepts, and draft text are processed during the Core Reasoning Layers.
- Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded commands within the user data.
- Capability inventory: The agent has the ability to execute shell commands (git) and perform network operations for updates.
- Sanitization: There is no mention of sanitizing or escaping the user-provided content before the agent evaluates it.
Audit Metadata