ginkida/agent-dispatch — MCP server + CLI that lets Claude Code agents delegate tasks to agents in other project directories.

Description and use

**Brief Summary**
ginkida/agent‑dispatch is a Python‑based MCP server and CLI that lets Claude Code agents hand off work to other agents living in separate project directories. It provides a full‑stack orchestration layer for parallel dispatch, session management, asynchronous jobs, and multi‑agent dialogues, turning ad‑hoc prompts and tools into repeatable, composable workflows.

**Value Proposition**

• **Workflow Automation** – Converts isolated prompts into deterministic pipelines, enabling teams to chain together specialized agents (e.g., data extraction → analysis → reporting) without writing custom glue code.
• **Scalable Orchestration** – Built‑in support for parallel execution, job queues, and persistent sessions lets you scale from a single developer experiment to enterprise‑level multi‑agent systems.
• **Standardized Memory & Tool Use** – Centralizes agent state and tool‑access patterns, reducing duplication and making it easier to audit and version‑control AI‑driven processes.

**Practical Adoption Path**

1. **Prototype** – Clone the repo, run the MCP server locally (agent-dispatch serve) and use the CLI (agent-dispatch run …) to connect an existing Claude Code agent to a test directory.
2. **Integrate** – Replace ad‑hoc script calls with the CLI or SDK calls (dispatch.submit_job(...)) in your CI/CD pipelines or internal tooling.
3. **Scale** – Deploy the MCP server in a container orchestration platform (Docker/K8s), configure persistent storage for session data, and enable the async job worker pool for high‑throughput workloads.
4. **Govern** – Add authentication (e.g., OAuth token) and monitoring hooks (Prometheus, logs) to meet security and observability requirements before moving to production.

**Production Readiness**

• **Activity & Community** – Recent commits (as of 2026‑06‑30), 28 stars, and a modest fork count indicate an active, albeit small, community.
• **Technical Maturity** – The project ships a stable API/SDK, a CLI, and clear Python type hints, covering the core orchestration features needed for production.
• **Risk Considerations** – License and security posture still need a formal review; there is only one known maintainer, so establishing a backup maintainer or contributing guidelines is advisable.

Overall, the project is mature enough for a serious pilot, especially in environments already using Claude Code agents and Python‑centric toolchains.

Problem

ginkida/agent-dispatch helps turn isolated prompts and tools into repeatable agent workflows.

Outcome

High for an OSS candidate: recent activity, adoption, and ecosystem signals are strong enough for a serious pilot.

Workflow

• coordinate multi-agent workflows
• add tool-use pipelines
• standardize agent memory

Evidence

MCP server + CLI that lets Claude Code agents delegate tasks to agents in other project directories. Multi-agent orchestration: parallel dispatch, sessions, async jobs, agent dialogues.

Source