The Agent Skills Directory

[COMMAND_EXECUTION]: The skill initiates execution of worker agents to perform tasks defined in a Directed Acyclic Graph (DAG) by calling the dispatch_workers function from the internal agent_factory module.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes task definitions and coordination messages from untrusted external sources. Ingestion points: The DAG structure is parsed from an external plan.md file, and task events are monitored via a file-system-based message bus (bus_path). Boundary markers: No explicit delimiters or instructions to ignore embedded malicious commands were identified in the orchestration logic. Capability inventory: The skill has the capability to spawn new agents, monitor their progress, and initiate recovery actions or retries. Sanitization: There is no evidence of validation, escaping, or sanitization applied to the task payloads or the DAG configuration before they are used to dispatch workers.

parallel-dispatch