The Agent Skills Directory

[SAFE]: The skill consists of instructional content for the Hatchet workflow engine. All referenced resources, such as the @hatchet-dev/typescript-sdk npm package, are official and well-known developer tools.
[EXTERNAL_DOWNLOADS]: Fetches the TypeScript SDK from the official npm registry and provides instructions for self-hosting Hatchet via its official GitHub-hosted Docker configuration.
[COMMAND_EXECUTION]: Includes standard shell commands for installing dependencies and managing Docker containers, which are necessary for the skill's stated purpose.
[PROMPT_INJECTION]: The skill facilitates processing external data via workflow steps. This creates a surface for indirect prompt injection if user-provided inputs are not properly sanitized.
Ingestion points: ctx.input() and ctx.stepOutput() in SKILL.md
Boundary markers: Absent
Capability inventory: Network requests (fetch), email dispatch (sendEmail), and database operations (createWorkspace) in SKILL.md
Sanitization: Absent in the provided code examples

hatchet