Mailbox MCP
Roamer MCP
FunParty
Mantra Meditation Timer
Twick.ing
Commune
Mailbox MCP
Roamer MCP
FunParty
Mantra Meditation Timer
Twick.ing
Commune
Mailbox MCP
Roamer MCP
FunParty
Mantra Meditation Timer
Twick.ing
Commune
Mailbox MCP
Roamer MCP
FunParty
Mantra Meditation Timer
Twick.ing
CommuneAbout Relay
Relay supervises coding agents through long-running tasks, recovers interrupted sessions, runs local checks, and produces signed, verifiable receipts.
Relay is an AI harness for coding agents that is built for work lasting longer than a single chat or terminal session. It supervises coding agents while you are away, keeps track of the task state, and helps a run continue after interruptions or recoverable failures.
Long-running agent work often breaks for ordinary reasons: a terminal disconnects, a process exits, a tool returns an unexpected error, or the session loses the exact context needed to continue. Relay is designed around that reality. It records the working context, remembers the settings used for the run, and resumes from the last known state instead of forcing you to reconstruct the task manually.
Relay runs directly on your machine and works with your existing repository and toolchain. It reads the conventions already present in the project, uses the commands and checks your repository expects, and does not require you to move source code into a separate hosted development environment. You can keep using the coding agent, editor, tests, scripts, and review process that already fit your workflow.
Verification is part of the workflow rather than an afterthought. Relay can run the configured tests and checks for a task, retain the resulting evidence, and produce signed receipts that can be reopened and checked again. A receipt shows what was run and what evidence was captured. It does not make an unlimited claim that a change is correct, secure, or ready for production. Developers still inspect the code, interpret the results, and decide what should be merged or shipped.
Relay is useful for substantial refactors, multi-step feature work, repository maintenance, migrations, test-driven changes, and other coding tasks that may need to run unattended. Solo developers can step away without losing the thread of a task. Teams can use the same local conventions and verification expectations while keeping review responsibility with the people who own the codebase.
The goal is simple: let coding agents work for longer without turning the process into a black box. Relay keeps the work moving, recovers when it can, preserves the context needed to continue, and gives you concrete evidence to review when you return.
Comments
Similar products

QA Crow
1 shared tagA murder of crows for your bug backlog
AnyFrame
1 shared tagSandboxes for your AI agents

AgentPeek
1 shared tagClaude Code and Codex in your Mac notch
SuperHQ
1 shared tagRun AI coding agents in real microVM sandboxes

Rosentic
1 shared tagCatch when coding agents break each other before merge

Radar
1 shared tagThe missing open-source Kubernetes UI
