The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from files within the workspace to generate structured planning data.
Ingestion points: scripts/run.py reads data from outline/concept_graph.yml and output/TUTORIAL_SPEC.md using the load_yaml and load_source_tutorial_spec_data functions.
Capability inventory: The skill performs file read and write operations within the workspace; no network operations, subprocess executions, or privilege escalation patterns were identified.
Boundary markers: There are no explicit boundary markers or instructions to disregard embedded commands in the source data during processing.
Sanitization: The skill utilizes standard parsing libraries but does not appear to perform specific validation or sanitization of the content extracted from the YAML or Markdown files.

module-planner