The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes npx agentic-flow commands to initialize swarms and orchestrate tasks, which involves executing external CLI tools within the environment.
[EXTERNAL_DOWNLOADS]: The agentic-flow package is fetched from the npm registry via npx as part of the orchestration workflow. This is a standard dependency for the framework provided by the author.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing external task descriptions and goals.
Ingestion points: Untrusted strings are ingested through the --task CLI parameter and the goal property in the autoOrchestrate method documented in SKILL.md.
Boundary markers: Delimiters or instructions to ignore embedded commands are not used in the examples.
Capability inventory: The framework allows spawning agents, executing code, and accessing shared memory.
Sanitization: No explicit sanitization or validation of the task descriptions is implemented.

Swarm Orchestration