explore-new-capabilities
Pass
Audited by Gen Agent Trust Hub on Nov 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill allows the agent to verify the feasibility of new feature ideas by performing small disposable checks and observing code execution within the project environment. This is a functional requirement for a developer-oriented tool.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources, specifically community feedback from GitHub Issues and Pull Requests as described in
references/prior-work.md. This represents a potential attack surface if the public comments contain malicious instructions intended to influence the agent. The skill attempts to mitigate this by directing the agent to separate the user's task and desired outcome from the requested implementation. - [EXTERNAL_DOWNLOADS]: The skill utilizes the GitHub search and issue APIs to cross-reference proposed ideas with prior work. These interactions with well-known official services for research purposes are considered legitimate and safe.
Audit Metadata