sleight-of-hand
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is entirely instruction-based and does not include any scripts, binaries, or external downloads. It operates within the existing tools available to the agent.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided file content and edit instructions. It addresses this attack surface with robust procedural guardrails: Step 2 (Isolate) requires identifying the smallest surface area and stopping if ambiguous; Step 3 (Stage) mandates a before-and-after diff verification; and Step 5 calls out unresolved risks. These measures minimize the risk of the agent being tricked into making unintended changes via malicious file content.
Audit Metadata