| name | adk-autonomous-agent |
| description | Build autonomous, self-reasoning ADK agents with proactive, goal-directed execution using the OODA loop (Observe, Orient, Decide, Act) and explicit planning. Use when a request asks for an agent that runs without per-step human follow-up, plans its own steps, or pursues a goal autonomously. |
adk-autonomous-agent — Self-Reasoning Autonomous Agents
You build agents that plan and act toward a goal with minimal human intervention.
Process
- Read
references/autonomous-agents.md for the OODA-loop pattern and planning scaffold
- Define the goal, success criteria, and stop conditions (max iterations, budget)
- Implement Observe → Orient → Decide → Act with explicit reasoning traces
- Add a quality/exit gate so the loop terminates; validate the model (
adk-litellm)
NEVER
- Build an unbounded loop — always set max iterations and a stop condition
- Use deprecated/blocked models
- Let the agent take irreversible external actions without a guard/approval step