Dropstone is a terminal coding agent with a model-agnostic monthly release cycle, approval-gated execution, and an unusual pricing play built around re-basing to whichever open model stack performs best each month.
Grass gives AI coding agents a persistent VM with phone-based approvals, remote diff review, and always-on sessions so work keeps moving after your laptop goes to sleep.
Handler is an open-source control plane for AI coding agents that runs each agent in its own sandbox, explains edits as they happen, and gives teams a cleaner way to review, fork, and steer long-running coding sessions.
Pi Coding Agent is a minimal, extensible terminal harness for AI coding workflows. It keeps the core small, supports many model providers, and lets developers shape prompts, packages, extensions, and runtime patterns around their own process.
Crew44 is a local-first workspace that turns installed coding agents into a coordinated crew of specialists with role-based models, persistent project memory, and structured handoffs across the same codebase.
Replicas runs Claude Code, Codex, and similar coding agents in isolated cloud development environments so teams can hand off engineering work from GitHub, Slack, Linear, or the web and come back to a reviewable pull request.
Joanium is a local-first AI desktop workspace for developers and power users that combines multi-model chat, project-scoped context, scheduled agents, real integrations, and on-device control over keys and data.
Sverklo is a local-first MCP server for coding agents that gives Claude Code, Cursor, Windsurf, Codex CLI and similar tools repo memory, symbol intelligence, blast-radius analysis, and git-pinned decisions before they edit code.
TAKT is a YAML-driven orchestration system for coordinating coding agents, human checkpoints, and reusable workflow records across tools like Claude Code and Codex.
Baton is a desktop app for running multiple AI coding agents in parallel, each inside its own git-isolated workspace, with built-in search, diffs, notifications, and MCP support.
Claude Mem turns agent sessions into a persistent searchable archive, then compresses and feeds the relevant context back into future coding runs across multiple agent tools.
Open SWE is LangChain's open-source asynchronous coding agent for planning, coding, testing, and opening pull requests from developer workflows such as GitHub, Slack, and Linear.
Agen runs fully autonomous coding agents in the cloud, letting teams assign work from the browser and get back branch-safe changes, previews, and pipeline fixes without sitting inside an IDE session.
Helmor is a local-first workbench for orchestrating multiple coding agents across worktrees on your own machine, with planning, review, and merge flow built in.
RoBrain gives AI coding teams a shared memory layer that records decisions, rejected options, and contradictions so agents stop reopening the same architecture debates from scratch.
Open Agents is Vercel's open-source reference app for building background coding agents that run in cloud sandboxes instead of on a developer laptop.
ContextPool gives coding agents persistent memory across sessions so teams stop re-debugging the same issues and re-explaining the same repo decisions.
ClawTab is a macOS control center for running and monitoring large numbers of Claude Code, Codex, OpenCode, and shell sessions through tmux.
Agentmemory gives Claude Code, Codex, Cursor, and other coding agents a persistent memory layer so teams stop re-explaining codebase context across sessions.
Contextberg turns your screen activity, browser history, and agent transcripts into local memory that Claude Code, Cursor, and similar agents can query through MCP.
Kilo Code is an open-source AI coding agent for VS Code, JetBrains, and the CLI, built for teams that want local-first workflows and broad model choice.
Warp is an agentic development environment that combines an AI terminal with coding agents for faster developer workflows.
Axel is a native macOS task manager for AI coding agents, letting developers queue work, dispatch tasks to Claude, Codex, OpenCode, or Antigravity, and approve actions from one inbox.
Contral is a teaching layer for AI coding agents that explains code as it is written across IDEs and agent tools such as Cursor, Copilot, Claude Code, Codex, Windsurf, and Antigravity.