catalyst-deploy
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides strong security guidance, requiring the agent to perform dry-runs before applying configurations and to seek explicit user consent before deleting resources or modifying permissions.
- [SAFE]: Explicit instructions are included to protect credentials, forbidding the agent from outputting secret values or API tokens and recommending secure secret references instead.
- [COMMAND_EXECUTION]: The skill utilizes platform-specific tools like catalyst_apply and catalyst_delete_resource to manage infrastructure. These capabilities are managed through procedural safety checks like confirmation tokens and user agreement.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from the cloud platform's API (e.g., resource status and manifests), creating a surface for potential indirect prompt injection.
- Ingestion points: Resource attributes returned by Catalyst API tools.
- Boundary markers: None; the instructions do not specify delimiters for data retrieved from tools.
- Capability inventory: Platform orchestration via tool calls to catalyst_apply and catalyst_delete_resource.
- Sanitization: Instructions mandate manual review of deployment previews and verification of readiness states.
Audit Metadata