Orchestration Agent
The Orchestration Agent's role as the primary planning, coordination, and memory agent — responsibilities, current status, and planned improvements.
Role
The Orchestration Agent is the main control and coordination component in the home lab. It acts as the primary interface between user intent and system execution. It plans work, coordinates development, maintains memory, and delegates implementation tasks to the Code Execution Agent under controlled rules.
Status: Implemented (concept and primary development workflow operational)
Responsibilities
The current implementation covers:
| Responsibility | Status |
|---|---|
| Architecture design | Implemented |
| Workflow planning | Implemented |
| Prompt generation for the Code Execution Agent | Implemented |
| Review and verification of generated work | Implemented |
| Memory and context management | Partially implemented |
| Documentation and reporting | Implemented |
| Coordination with the Research Agent | Implemented |
| Coordination with the Media Agent | Implemented |
| Teaching and explanation | Implemented |
What the Orchestration Agent is
The Orchestration Agent is a control agent, not an execution agent. It does not write code directly. It designs solutions, generates prompts for the Code Execution Agent, reviews the returned output, and integrates it into the project. This separation keeps implementation isolated and reviewable.
The Orchestration Agent maintains context across a session: project state, file structure, recent changes, active constraints, and open decisions. This context informs all planning and prompt-writing work.
What the Orchestration Agent is not
The Orchestration Agent does not:
- Execute code directly in production or staging environments
- Interact with git (commits and pushes remain under direct user control)
- Make deployment decisions independently
- Run research missions directly (it initiates them via the Research Agent's task queue)
Memory and context
The current memory system is implemented as conversation context augmented by written notes and documentation files. A more structured memory review and distillation workflow is planned.
Relationship with the Code Execution Agent
The Code Execution Agent operates as the Orchestration Agent's implementation arm. The Orchestration Agent designs the solution and writes an explicit, scoped prompt. The Code Execution Agent executes the prompt and returns the result. The Orchestration Agent then reviews, integrates, and tests.
→ Code Execution Agent for the full step-by-step procedure.
Relationship with the Research Agent and Media Agent
The Orchestration Agent can initiate research missions by writing task files to the Research Agent's task queue. Research reports are then consumed by the Orchestration Agent or the user. The Orchestration Agent can also delegate media requests to the Media Agent.
The agents do not share a session or compete for resources.
Architecture note
Some capabilities of the Research Agent and Media Agent were initially prototyped in a shared runtime. The architecture models them as separate logical agents because their responsibilities, tools, and risk profiles are different.
Future work
- Control UI — browser-based interface for monitoring and managing the Orchestration Agent
- Structured job management — explicit job IDs, status tracking, and cancellation
- Memory review workflow — scheduled memory distillation and knowledge base maintenance
- Structured reporting — standardized output formats for development sessions
- Safer operational boundaries — clearer rules for autonomous vs. supervised operations
Related pages
→ Code Execution Agent — implementation workflow → Research Agent — research mission delegation → System Overview — full architecture context → Data Flow — how requests and results move through the system