Open Source Coding Agents
Coding agents use language models to interpret development tasks and take actions in a software project. Depending on their permissions and design, they can inspect files, explain code, make edits, run tests or shell commands, and continue through multiple steps. They help automate routine work and support exploration of unfamiliar codebases, while requiring review because generated changes and commands may be incorrect or unsafe.
Open source options include command-line assistants, desktop applications, agent runtimes, and tools for coordinating sessions or models. When choosing one, consider its license, maintenance activity, model compatibility, data handling, access controls, and requirements for running it. Check how it integrates with your editor, terminal, version control, and existing workflow. These tools may suit individual developers, teams, and people building or studying agent-based development systems.
109 repositories · updated October 4, 2026
agent-observability: Monitor AI Coding Agents Locally
A self-hosted OpenTelemetry stack for monitoring Claude Code and OpenAI Codex. It routes telemetry to Prometheus, Loki, and Tempo, then presents usage, performance, and activity in Grafana dashboards.
OrcaReplay: Record and Replay AI Agent Runs
OrcaReplay records AI agent runs so developers can inspect what happened, replay interactions offline, and fork a run from a checkpoint onto another model. It is a TypeScript CLI for debugging and comparing agent behavior without modifying the agent.
prime-agent: Run Self-Improving Agents for Coding Work
Prime Agent is a coding and research agent built for long-running tasks, using a persistent Python environment, recursive subagents, and durable working context. It suits developers and researchers who need agents to keep working across sessions.
codeg: Coordinate AI Coding Agents in One Workspace
Codeg brings multiple AI coding agents into a shared workspace for conversations, code changes, and delegated tasks. It suits developers who want to compare or coordinate agents across desktop, self-hosted, and Docker setups.
vicoa: Coordinate Coding Agents Across Devices
Vicoa is a self-hostable command center for running coding agents on local and remote machines. It brings agent sessions, task management, and review tools to desktop, web, mobile, and CLI clients.
ai-session-search: Search Local AI Coding Sessions
ai-session-search indexes local transcripts from multiple AI coding tools and lets you search sessions, messages, and edited files. Use it from a Rust-powered CLI, MCP server, Rust library, or Python API.
pydantic-ai-harness: Add Capabilities for Long-Running AI Agents
A Python capability library that extends Pydantic AI agents with tools and support for coding, research, memory, delegation, and context management. This repository is no longer updated; development continues in the Pydantic AI repository.
open-ace: Govern AI Coding Agents in a Self-Hosted Workspace
Open ACE brings AI coding CLIs into a self-hosted workspace with remote execution and administrative controls for keys, access, quotas, usage, and audits. It suits engineering teams that need to manage agents across private machines and shared workflows.
spec-kitty: Govern AI Coding Workflows with Specs
Spec Kitty is a Python CLI that keeps specifications, tasks, review decisions, and merge state in Git while coordinating AI coding agents. It suits teams that need repeatable, traceable workflows and parallel implementation, but may be excessive for small one-off changes.
Kun: Local-First AI Agent Workspace
Kun is a local-first workspace for using AI agents across coding, writing, design, research, and automation. Its desktop GUI and terminal TUI share one runtime, keeping tasks and approvals connected.
fx: Run a Unix-Style Coding Agent in Your Terminal
fx is a native Zig coding-agent CLI with an interactive shell, one-shot requests, and embedding options. It suits developers who want an extensible, model-agnostic agent that fits terminal workflows rather than a full-screen IDE.
lat.md: Keep Codebase Knowledge in a Linked Markdown Graph
lat.md is a TypeScript CLI and markdown-based knowledge graph for documenting codebase concepts and linking them to source code. It helps coding agents and developers find context, keep references consistent, and retain decisions across work sessions.