The Agent Skills Directory

[PROMPT_INJECTION]: The skill's primary function is to ingest and synthesize information from external web research and subagent outputs, creating an indirect prompt injection surface where malicious data could attempt to influence the agent's behavior.
Ingestion points: External data enters the context through research subagents and codebase exploration tasks described in SKILL.md and references/delegation-patterns.md.
Boundary markers: The skill utilizes structured 'prompt packets' defined in references/prompt-packets.md to define scopes for sub-tasks.
Capability inventory: The orchestrator has the capability to spawn sub-processes and tasks that can modify files or access the network, as outlined in references/harness-adapters.md.
Sanitization: The references/synthesis-gate.md provides a manual verification framework for the agent to validate claims against direct evidence before finalizing outputs.
[COMMAND_EXECUTION]: The instructions direct the agent to utilize native platform tools to spawn and manage sub-agents or tasks.
Evidence: references/harness-adapters.md contains logic for interacting with various agent harnesses (such as Codex and Claude Code) to perform delegation.

parallel-subagents