workflow-init
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- External Content Retrieval: The skill directs the agent to fetch installation and configuration instructions from an external domain.\n
- Evidence: URLs under
https://workflow-sdk.dev/docs/getting-started/are fetched based on the detected framework.\n- Indirect Prompt Injection Surface: Retrieving and following step-by-step guides from an external data source creates an ingestion point where online content could influence agent actions.\n - Ingestion points: External documentation URLs fetched in Step 2 of
SKILL.md.\n - Boundary markers: Absent. There are no explicit boundaries or instructions to ignore potential command overrides within the fetched text.\n
- Capability inventory: The skill permits the agent to inspect
package.json, install packages, create files, and execute commands.\n - Sanitization: Absent. The content is followed directly without pre-validation.\n- Command Execution: The workflow setup involves running development tools and validation scripts.\n
- Evidence: Step 3 describes running dev servers, executing
curlcommands, and invokingnpx workflow.
Audit Metadata