What changed
The Agentic System Lab has released Harness Relay, a local group chat application that integrates a variety of coding CLIs and open-source AI agents. The project supports popular tools such as Claude Code, Codex, Grok, Cursor, Antigravity, Copilot, and Muse Code CLIs. Additionally, it incorporates five open-source coding agents: OpenCode, Goose, Cline, OpenHands, and Pi. Users can select their preferred model backends for these agents. The application allows for the assignment of specific roles to each agent and the configuration of their turn-taking behavior, enabling them to collaborate on code-related tasks or answer questions collectively.
The project is primarily written in TypeScript and has a recent release dated November 14, 2026. The repository indicates a strong focus on agent orchestration and agent-client protocols, with topics including agent-orchestration, agentic-coding, ai-agents, and multi-agent systems. It also highlights features like local-first and self-hosted operation, suggesting a focus on privacy and control for developers.
Why it matters for builders
Harness Relay provides a unified environment for developers to interact with and manage multiple AI coding assistants simultaneously. This capability is particularly valuable for complex projects where different agents might excel at specific tasks, such as code generation, debugging, or documentation. The ability to define roles and turn-taking sequences allows for fine-grained control over AI collaboration, potentially leading to more efficient and effective code development workflows. Builders can leverage this tool to experiment with advanced multi-agent AI strategies directly within their local development setup.
Practical impact
Developers can utilize Harness Relay to set up a collaborative coding environment where various AI agents and CLIs work together. This could involve assigning one agent to write initial code, another to review and refactor it, and a third to generate documentation, all within a single chat interface. The project's support for multiple model backends and its self-hosted nature offer flexibility and customization. The latest release, dated November 14, 2026, suggests ongoing development and potential for future enhancements. Builders interested in exploring advanced AI-driven coding assistance can explore this project for its unique approach to agent orchestration.
Caveats and source limits
The provided repository metadata is limited in detail regarding specific usage instructions, architectural diagrams, or performance benchmarks. The readme_summary mentions "6 AI signals" and "4 developer signals" but does not elaborate on what these specific signals entail. Information on system requirements, installation procedures beyond the presence of a package.json file, and the exact nature of the "agent-client-protocol" are not detailed. The license is listed as "other," which may require further investigation for commercial or distribution use. The project has a relatively low star count (27) and fork count (4), indicating it is in an early stage of adoption or development.
Sources
Claim check: 9/9 supported claims - 9 evidence links - 100% avg confidence
- Harness Relay is a local group chat application for coding CLIs and open-source coding agents.supported - github.com
- It supports CLIs like Claude Code, Codex, Grok, Cursor, Antigravity, Copilot, and Muse Code.supported - github.com
- It integrates five open-source coding agents: OpenCode, Goose, Cline, OpenHands, and Pi.supported - github.com
- Users can choose model backends, assign roles, and configure turn-taking for agents.supported - github.com
- The project is written in TypeScript.supported - github.com
- The latest release was on November 14, 2026.supported - github.com
- The repository has 27 stars.supported - github.com
- The repository has 4 forks.supported - github.com
- The project emphasizes local-first and self-hosted operation.supported - github.com
Caveats
- Single-source caution: verify critical details at the linked source.
Radar score 87/100 - how it was calculated
- Reliability 82: GitHub metadata supports source trust
- Freshness 100: Fresh GitHub release date
- Novelty 77: Fresh GitHub release
- Technical 81: Repository technical metadata
- Developer 96: Developer tooling signals
- Ecosystem 72: Fresh GitHub release
- Confidence 96: Claims have reliable evidence
Discussion
Loading comments...