The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from clarified-requirements/index.md which creates a surface for indirect prompt injection if the source requirements are untrusted.
Ingestion points: The agent reads requirements from the clarified-requirements/ directory.
Boundary markers: The skill does not provide instructions to use delimiters or ignore embedded commands within the input requirement documents.
Capability inventory: The agent is granted Read, Write, and Bash permissions; it uses a local Python script scripts/parse_plan.py to validate generated plans.
Sanitization: No specific sanitization or structural validation is applied to the input requirement text before it is processed by the agent.

testcase-planner