LightfieldAI-native CRM that builds itself and does work for you
Promoted
Agent Studio
Your local hub for agent-powered development
9 followers
Your local hub for agent-powered development
9 followers
✨ Agent Studio allows you to run autonomous coding agents without turning your workflow into a black box. Customize agents for each task, curate project knowledge in a built-in context library, switch smoothly between workspaces, and follow the entire pipeline as agents plan, execute, ask questions, generate artifacts, request approval, and prepare final work for review. It brings context, agents, approvals, and project history into one local workspace hub for teams.
Free Options
Launch Team / Built With
Hey PH 👋 Justen here, building Agent Studio.
What is Agent Studio?
Agent Studio is a local-first desktop control plane for running coding agents like Codex, Claude Code, and even open source models with more visibility, structure, and safety than a terminal alone.
❓The Problem
Coding agents are powerful, but the terminal experience is still too opaque. You ask for a bigger change, then watch a stream of logs and hope the agent understood the task, touched the right files, ran the right checks, and didn’t make risky changes. Once you start using multiple agents and providers, it gets even harder to track what happened, what needs review, and when it’s safe to apply the work.
☑️ The Solution
Agent Studio turns agent work into a visible, staged workflow. Open a repo, run studio, create a task, attach context, and let the Studio Pipeline move through orchestration, planning, execution, review, and reporting. You can watch every stage live, approve plans, answer questions, inspect artifacts, open the completed worktree in your IDE, and manually merge/apply changes only when you’re ready.
💡 What makes it different
Agent Studio is not another chat app or CLI. It is a local desktop dashboard for supervising real coding-agent work in your own repositories. State lives under ~/.studio, execution happens against local git worktrees, and humans stay in control through explicit approval gates, questions, provider health checks, and final merge/apply actions.
Key Features
⚡ Visual Studio Pipeline: orchestrator, planner, executor/workers, review, and reporter stages you can inspect live
🔍 Real-time visibility: activity feeds, logs, tool output, node details, structured plans, reviews, reports, approvals, and questions
🌳 Safer git workflows: isolated worktrees, integration worktree, conflict visibility, Open in IDE, and manual Merge / Apply
🧠 Better context for agents: attach markdown docs, uploaded files, screenshots, URLs, one-off text, and selected project directories
🧩 Reusable agents: save role-specific agents with prompts, tags, resources, provider/model settings, effort, and permission profiles
🔒 Human-in-the-loop controls: plan approval, review gates, permission prompts, clarification questions, resume/restart controls, and minimized-interruption mode
🩺 Provider readiness: Bespoke Studio Engine/model, Codex, and Claude Code support with diagnostics, version checks, capabilities, fix hints, and studio doctor
🙋🏻 Who It’s For: Developers, founders, and builders already using CLI coding agents who want to delegate bigger engineering tasks without losing the audit trail. If you use Codex, Claude Code, or local agent workflows and have ever thought “I need to see what this thing is actually doing before I merge it,” Agent Studio is for you.
❤️ What We’d Love From You: Try it on a real repo and tell us: where do you most want more control — planning, context, reviews, provider selection, or merging? We’ll be reading every comment and using your feedback to shape the next version. 🙏