The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and act upon data from external project sources such as ticket files and specifications, which could contain instructions intended to influence agent behavior.
Ingestion points: Reads work items from the .scratch/ directory and project documentation (e.g., SKILL.md Core Loop Step 1 and WORKFLOW.md Step 1).
Boundary markers: The skill lacks explicit boundary markers or directives to the agent to ignore embedded instructions within the ingested ticket content.
Capability inventory: The workflow allows for the execution of shell commands (e.g., npm test in WORKFLOW.md Step 5) and the invocation of additional model-driven skills such as light-tdd and light-research.
Sanitization: No input validation, sanitization, or filtering of the ticket content is described before it is processed by the agent.

light-implement