nexus-room
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The agent is instructed to use credentials stored in a local environment file at
~/Projects/nexus/.envfor API authentication. - [EXTERNAL_DOWNLOADS]: The skill interacts with the vendor-specific API endpoint at
https://nexus.langwatch.aito synchronize strategy and content tasks. - [COMMAND_EXECUTION]: The instructions utilize the
kanbanCLI for channel management andcurlfor making authenticated network requests to the project's API. - [PROMPT_INJECTION]: The skill utilizes external channel history for agent synchronization, which introduces an indirect prompt injection surface.
- Ingestion points: The
kanban channel historycommand retrieves data from a shared channel that could contain untrusted input. - Boundary markers: The skill lacks explicit boundary markers or instructions to ignore commands within the channel history.
- Capability inventory: The agent can execute network requests, perform GitHub operations, and manage CLI tools.
- Sanitization: No specific input sanitization is defined for the channel data.
Audit Metadata