The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes the mthds-agent CLI tool throughout the build process to perform validation, conversion, and assembly tasks. Examples include mthds-agent pipelex concept for schema validation and mthds-agent pipelex assemble for bundle creation in the mthds-wip/ directory. These operations are consistent with the vendor's (mthds-ai) documented infrastructure.
[PROMPT_INJECTION]: The skill incorporates an indirect prompt injection surface as it processes user requirements into pipeline instructions.
Ingestion points: Untrusted data enters the context through user-defined input concepts (Phase 3) and processing transformations (Phase 5).
Boundary markers: The skill explicitly uses @variable for delimited block insertion and $variable for inline interpolation as defined in the 'Quick Reference' section.
Capability inventory: Capabilities include shell command execution via the mthds-agent binary and remote model inference via PipeLLM operators.
Sanitization: The process includes a mandatory validation step (mthds-agent pipelex validate) in Phase 9 to verify bundle integrity prior to execution.

build