The Agent Skills Directory

[SAFE]: The skill serves as a reference guide and boilerplate repository for agent orchestration. The provided Python and TypeScript scripts implement standard ReAct and multi-agent patterns using established libraries such as OpenAI, CrewAI, and LangGraph.
[CREDENTIALS_UNSAFE]: No hardcoded secrets were found. The code templates correctly utilize environment variables like OPENAI_API_KEY and DATABASE_URL for configuration. Example connection strings in documentation use local placeholders for development purposes.
[COMMAND_EXECUTION]: The coordination script uses subprocess.Popen to manage multiple parallel instances of the agent platform CLI for testing scenarios. This behavior is documented and restricted to launching the specific instances required for the skill's multi-scenario orchestration feature.
[INDIRECT_PROMPT_INJECTION]: While the agent templates process user-supplied tasks, the skill proactively includes examples of safety guardrails (such as PII detection and content safety filters) and recommends boundary management in its architectural instructions.

agent-orchestration