agentfield

AgentField is an open-source control plane written in Go that treats AI agents as first-class backend services, exposing each agent function as a REST endpoint with routing, async execution, memory, and cryptographic audit trails baked in. Developers write agent logic in Python, Go, or TypeScript using an SDK, then AgentField handles production concerns including canary deploys, multi-agent coordination, and human-in-the-loop approval flows via webhook-triggered execution pauses. It integrates directly with Claude through the Anthropic API, specifying models such as claude-sonnet-4, and its Harness orchestration layer supports multi-turn coding workflows inside Claude Code alongside other coding agents. A single slash command, /agentfield, typed into a coding agent generates a complete Docker Compose stack with a wired control plane and a callable REST endpoint. Every agent receives a cryptographic identity, and every decision produces a traceable audit trail, making the project most relevant to engineering teams deploying multi-agent systems in Kubernetes environments where observability and accountability are operational requirements.

Deploy agentfield to your cloud

Ship this repo to production in minutes. Each platform spins up its own environment with editable env vars.

Maintain this repo? Add a badge to your README

Drop the badge into your GitHub README to show it's tracked on ClaudeWave. Each badge links back to this page and reflects the live Trust Score.

Markdown (README)Copy

[![Featured on ClaudeWave](https://claudewave.com/api/badge/agent-field-agentfield)](https://claudewave.com/repo/agent-field-agentfield)

HTMLCopy

<a href="https://claudewave.com/repo/agent-field-agentfield"><img src="https://claudewave.com/api/badge/agent-field-agentfield" alt="Featured on ClaudeWave: Agent-Field/agentfield" width="320" height="64" /></a>