The Agent Skills Directory

[SAFE]: The skill provides documentation and reference examples for implementing a workflow plugin. It follows standard software engineering practices for CommandKit-based projects.
[SAFE]: No obfuscation, data exfiltration, or unauthorized command execution patterns were found in the instructions or code snippets.
[INDIRECT_PROMPT_INJECTION]: The skill includes patterns for handling external inputs (e.g., user IDs from interactions) to trigger workflows. While this represents a data ingestion surface, the provided code snippets only perform benign actions such as sending messages through a client, and the documentation explicitly advises validating external inputs and permission boundaries as a best practice.
Ingestion points: userId parameter in greetUserWorkflow (referenced in references/02-workflow-and-step.md) and ctx.interaction data (referenced in references/03-start-workflow.md).
Boundary markers: None explicitly used in the string templates of the examples.
Capability inventory: Discord client interactions (client.users.fetch, user.send) and workflow lifecycle management (workflow.start).
Sanitization: Included in 'Best practices' sections across all reference files, which recommend validating external inputs before side effects.

commandkit-workflow