The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill describes a framework for processing external data via workflows. While this presents a surface for indirect prompt injection, the documentation explicitly recommends and demonstrates using Zod schemas for input validation to mitigate risks.\n
Ingestion points: workflow() input and step.run() arguments (e.g., in references/sonamu-integration.md).\n
Boundary markers: Examples show the use of z.object() and z.url() for input validation.\n
Capability inventory: Workflows can execute logic, perform database operations, and log data.\n
Sanitization: Recommends Zod schema validation and explicit validation of untrusted manual input.\n- [COMMAND_EXECUTION]: Instructions include standard shell commands for managing the development environment and package installation.\n
Evidence: pnpm add, pnpm sonamu sync, and pnpm sonamu dev.\n- [EXTERNAL_DOWNLOADS]: The skill directs the user to install standard Node.js packages from the NPM registry.\n
Dependencies: @sonamu-kit/tasks, @logtape/logtape, and knex.

sonamu-tasks