The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a local Python validation script (scripts/validate_cascade.py) to verify the structure of workflow documents. This is a standard utility function provided as part of the skill.
[INDIRECT_PROMPT_INJECTION]: The skill implements a workflow where the agent's actions are driven by content read from local markdown files in the .cascade/ directory. This creates a surface where the agent ingests data that influences its next steps.
Ingestion points: Chain documents and boundary receipts are read from the .cascade/ directory during PLAN, ADVANCE, and TAKEOVER modes.
Boundary markers: The skill uses YAML frontmatter and specific Markdown headers (e.g., ## Current, ## Authority and budgets) to structure and delimit instructions.
Capability inventory: The agent is empowered to perform state changes, file writes, and other commands as explicitly authorized in the chain's authority field.
Sanitization: A structural validation script (scripts/validate_cascade.py) is used to ensure file integrity, although semantic safety relies on the agent's judgment and the user-defined authority grants.

cascade