wisp GitHub

v0.15.0 · free and open source

Send coding agents off. Read what they did.

macOS · Linux and Windows in beta · Node 20+ · MIT

Hand tasks to Claude Code, Codex, Cursor and four other agent CLIs. They run in the background, each on its own branch. You get a one-screen report when one is done, or a question when it's stuck. Never a transcript.

✦AutoBalanced Worktree Try it: type a task, press ↵

Needs you 0

Working 0

Runs the agent CLIs you already pay for

A morning with wisp

Three tasks, ninety minutes,
about four minutes of your attention.

  1. Hand off three tasks.

    A flaky test, a Stripe migration, a Node upgrade. Each goes to its own worktree and branch, so they can't collide. Then you go and do something else.

  2. One of them has a question.

    Claude Code found a decision that's yours to make. It pauses and you get a notification. Pick an option and the answer goes straight back into the running session.

  3. A report lands.

    Not a log. A headline, the numbers that changed, the real diff, the test run, and what to do next. You can read it in a minute.

  4. Ship it.

    wisp has already drafted the commit message and PR description. One click commits, pushes and opens the pull request.

  5. Merged. Queue clear.

    The card follows CI and review on its own. Merge squash-merges and fast-forwards your local main. Two more reports are waiting when you're ready.

wisp
Three tasks in the Working group: fixing a flaky test, a Stripe webhook migration and a Node 22 upgrade, each with a live status line. The Stripe migration task paused on a question with three numbered options. A finished report for the flaky test: headline, stats and the start of the diff. The Ship it dialog with a drafted commit message and pull request description. The queue later: two reports in Needs you, and the flaky-test fix merged as PR #186.

The report

Same run. Pick what you'd rather read.

Fix the flaky session-expiry test
A finished task expanded: headline 'Session-expiry test no longer flakes', three stats, a diff of src/auth/session.ts, a green test run, a file list, a callout and next steps.
  • Real diffs. Captured from the agent's own edits, not summarised. Revert one file, or roll back the whole turn.
  • Real tests. The command it ran and the result. Re-run it from the card.
  • Next steps. One click turns a suggestion into a follow-up in the same session.
  • Your call. E marks it done and opens the next one. T takes it over in your terminal.

Compare

Not sure which agent?
Run both and keep the winner.

Click vs in the launcher and one prompt runs on two harnesses at once, each in its own worktree. wisp lines them up: time, cost, files, tests, and both reports. Keep this archives the other.

Compare view: Claude Code finished in 16 minutes for $0.94 with 341 tests passing; Codex finished in 11 minutes for $0.38 with 2 failing tests. Each column has a Keep this button.

You stay in control

Pick the agent. Set the leash.
Take the wheel whenever you like.

The launcher's harness picker: Auto, Claude Code, Codex, opencode, Cursor, Antigravity and Grok, with Balanced, Cheap and Best goals.
Auto, or your pick. Choose a harness, model and thinking level per task, or let Auto choose from what worked before. It learns from what you merge, your 👍/👎 and how many follow-ups a run needed.
The Automations dialog listing three recipes (release, deps-audit, triage) with Run buttons.
Recipes and triggers. Save the prompts you keep retyping; {version} becomes a field. Run them on a schedule, when CI fails on main, or when an issue gets a label.
Plan

Read-only. It investigates and proposes; nothing on disk changes. Issue triggers always start here.

Edits

File edits are accepted; anything else asks. The default.

Full auto

No prompts. Fine in a worktree with checkpoints before every turn.

Press T to take over.

wisp opens your terminal in the task's worktree with the harness's own resume command. Ghostty, iTerm, Warp, kitty, WezTerm, Terminal, Windows Terminal and more.

HarnessTake over runs
Claude Codeclaude --resume <id>
Codexcodex resume <id>
Cursorcursor-agent --resume <id>
opencodeopencode --session <id>
Antigravityagy --conversation <id>
Grokgrok --resume <id>

From the terminal

Or never leave your shell.

wisp "…" launches a task in the current folder and starts the app if it isn't running. --wait blocks until the run is done and prints the report as markdown, so scripts, CI and other agents can use wisp too.

Also: an MCP server (wisp mcp), an agent skill, a JS SDK, an HTTP API and webhooks. Integrations →

~/code/api-server

        

      

    


    
    

      
Also in the box

      

        
Machines
Run agents on a dev box or cloud VM over SSH. One prompt can go to this computer and several machines at once.

        
Subagents
Give the lead a team on other harnesses. Claude Code can hand tests to three Codex copies through a delegate tool.

        
Checkpoints
Every turn is snapshotted before it starts. Restore to before any turn, or revert one file.

        
It learns
After a session, a short retro records your preferences and the agent's mistakes. Later runs get them in their prompt.

        
Phone
Questions and finished tasks go to ntfy or Telegram. Answer from there, or start a task with /new.

        
Keyboard
N new, J/K move, O open, E done and next, ⌘K search everything.

        
Attachments
Paste a screenshot, drop a screen recording, a log or a PDF into the launcher.

        
Usage
Spend, success rate and questions asked, per project and per harness and model.

        
Small and local
A Node server with zero runtime dependencies and a vanilla JS UI. State is plain JSON in ~/.wisp.

      

    


    
    

      

        
        
wisp Cloud · early preview

        
Let it run while you sleep.

        
Managed machines, queues that keep going after you close the laptop, and one dashboard for your team. wisp on your computer stays free.

        

          Join the Cloud waitlist
          See what's coming
        

      

    


    
    

      

        
Pick the task you've been putting off.

        
A flaky test, a dependency bump, a small bug. You'll be set up in about a minute.

        
    
          
  1. 1Node 20+ and one agent CLI, signed in
    Claude Code, Codex, Cursor CLI, opencode, Antigravity CLI or Grok Build.
    2. 
          
  2. 2Run wisp
    Opens the desktop app. Add --browser for localhost:4477.
    3. 
          
  3. 3Describe it, pick the folder, press Enter
    Then go do something else. wisp tells you when it's done or needs you.
    4. 
        

      

    


    
    

      
Questions

      

        
What is wisp?
wisp is a free, open-source launcher for AI coding agents. It runs agents like Claude Code and Codex on your projects in the background. When one finishes, it hands you a short report (the diff, test results, commits and next steps) instead of a long transcript.

        
Which coding agents does wisp support?
Claude Code (claude), OpenAI Codex (codex), opencode, Cursor CLI (cursor-agent), Antigravity CLI (agy) and Grok Build (grok). Each harness is one adapter file, so adding another is small.

        
Is wisp free?
Yes. wisp is MIT-licensed and free. You use your existing Claude Code, Codex, opencode, Cursor, Antigravity or Grok account, and those providers bill as usual.

        
Does my code leave my machine?
wisp itself is a local server. Sessions and state are plain JSON in ~/.wisp. Your code only goes to the agent provider you choose, same as running that CLI directly. wisp sends anonymous usage stats with no prompts, paths or agent output. Turn them off with WISP_TELEMETRY=0 or DO_NOT_TRACK=1.

        
Can I take over a session in my terminal?
Yes. Press T and wisp opens your terminal in the project with the agent's resume command, for example claude --resume <id>. It works with Terminal, iTerm, Ghostty, Warp, kitty, WezTerm, Windows Terminal and more.

        
Can other tools send tasks to wisp?
Yes. wisp ships an MCP server (wisp mcp), an agent skill, a scriptable CLI, a JS SDK, an OpenAPI-described HTTP API and webhooks. You can hand tasks to it from other agents, editors, scripts or CI.

        
How do I install wisp?
Install Node.js 20 or later and sign in to at least one supported agent CLI. Then run npx wisp-agents to open the desktop app, or npx wisp-agents "fix the flaky auth test" to launch a task in the current folder.

      

    


    
    

      

        
        
Tomorrow morning, start here.

        

          
          Read the code on GitHub