chatcut-editor-handoff
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external data from ChatCut tool results, which could potentially be influenced by attacker-controlled content in external project data.
- Ingestion points: The skill ingests
browserHandoffobjects and tool return values as described inSKILL.mdto determine navigation targets. - Boundary markers: The instructions provide clear constraints on how to handle the data, directing the agent to use specific fields rather than free-text interpretation.
- Capability inventory: The skill utilizes the
InAppBrowsertool to perform navigation and session management. - Sanitization: The skill enforces "Link hygiene" by requiring the agent to strip sensitive authentication parameters (
editor-boot-token) from URLs before they are presented to the user in the UI, effectively mitigating data exposure risks. It also warns against hand-building or rewriting URLs to maintain integrity.
Audit Metadata