The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The task delegation pattern in SKILL.md involves passing context to sub-agents, creating a surface for indirect prompt injection.\n- Ingestion points: Data enters the agent context through workspace paths, context, and context_summary fields as described in the execution instructions within SKILL.md.\n- Boundary markers: Instructions in SKILL.md mention scoping tasks with concrete deliverables and acceptance criteria, but omit instructions for explicit markers to delimit or ignore embedded instructions in the passed context.\n- Capability inventory: SKILL.md specifies that the analyst sub-agent profile has access to the session Code Interpreter for data inspection and computation.\n- Sanitization: No sanitization, escaping, or filtering of the interpolated context data is described in the SKILL.md instructions.

delegate-work