The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to run a local static analysis script (scripts/validate-workflow.mjs) to verify generated workflow code before finalization.
[INDIRECT_PROMPT_INJECTION]: The skill handles untrusted data via the args object in generated workflows. It provides the following evidence chain: 1. Ingestion points: The args global variable in the JavaScript body of the workflows. 2. Boundary markers: The skill recommends using JSON.stringify to pass data between agents. 3. Capability inventory: Subagents triggered by the agent() call have access to the platform's standard tools (bash, filesystem). 4. Sanitization: The skill provides mandatory boilerplate code for input normalization and requires the use of JSON schemas to enforce structured output validation.

workflow-creator