# wisp > Your coding agents, as code. wisp runs Claude Code, Codex, Cursor and every other coding agent from your repo, in parallel, on any of your computers or in the cloud, and keeps every task they did next to the code they wrote. Each page below is plain markdown. `llms-full.txt` has all of them in one file. ## Get started - [Documentation](https://trywisp.dev/docs/index.md): Manage the code your agents write. Run Claude Code, Codex and every coding agent from the app, the TUI or the CLI, declare them as code, and track every line they write. - [Install wisp](https://trywisp.dev/docs/start/install.md): Get wisp on macOS, Linux or Windows, with npm, Homebrew, the install script or the desktop app. - [Your first task](https://trywisp.dev/docs/start/first-task.md): Start a task from the desktop or phone app, the terminal UI or the command line, watch it work, follow up, and ship what it wrote as a pull request. - [Your repo's .wisp folder](https://trywisp.dev/docs/start/your-repo.md): Give your repo a .wisp/ folder so its agent setup and the tasks that built it live next to the code. ## Concepts - [Declare, Run, Track](https://trywisp.dev/docs/concepts/declare-run-track.md): The three things wisp does, and how the rest of these docs map onto them. - [Tasks](https://trywisp.dev/docs/concepts/tasks.md): What a task is, its turns, its modes, its worktree, its answer, and how it ends. - [Agents and harnesses](https://trywisp.dev/docs/concepts/agents.md): Harnesses are the coding agent CLIs wisp runs; agent profiles are your named ways of running them. - [Crew](https://trywisp.dev/docs/concepts/crew.md): Long-lived agents with a role, a brief, a notebook and a schedule, that you hand jobs to. - [Formulas (your setup as code)](https://trywisp.dev/docs/concepts/formulas.md): A folder of .wisp.yml files that declares agents, crew, machines, skills, MCP servers, rules and workflows. - [The .wisp folder](https://trywisp.dev/docs/concepts/wisp-folder.md): What a repo's .wisp/ folder holds, and how each task in it links to the code it wrote. - [Machines and wisp Cloud](https://trywisp.dev/docs/concepts/machines-and-cloud.md): Run tasks on your other computers and on a server, and reach your wisp from anywhere. - [Plugins](https://trywisp.dev/docs/concepts/plugins.md): Everything in wisp is a plugin, the built-in parts included, with the same API and permissions as yours. - [Factories](https://trywisp.dev/docs/concepts/factories.md): Run a task through stations (spec, build, checks, review, ship) instead of one agent, looped back to build until it passes. ## Guides - [Guides](https://trywisp.dev/docs/guides.md): How to do each thing with wisp, from the app, the terminal UI and the command line. - [Start with Claude Code](https://trywisp.dev/docs/guides/agents/claude-code.md): Run Claude Code in wisp on your own Claude plan or key. Pick its model and effort, add more accounts, and open a task back in Claude Code. - [Start with Codex](https://trywisp.dev/docs/guides/agents/codex.md): Run OpenAI's Codex in wisp on your own ChatGPT plan or API key. Pick its model and effort, add more accounts, and open a task back in Codex. - [Start with Cursor](https://trywisp.dev/docs/guides/agents/cursor.md): Run the Cursor CLI (cursor-agent) in wisp on your own Cursor plan or key, pick its model, and open a task back in Cursor. - [Start with opencode](https://trywisp.dev/docs/guides/agents/opencode.md): Run opencode in wisp, with free models and no account, or on any provider you sign in to. Pick its model and variant, and open a task back in opencode. - [Start with GitHub Copilot](https://trywisp.dev/docs/guides/agents/copilot.md): Run the GitHub Copilot CLI in wisp on your own Copilot plan. Pick its model and effort, and open a task back in Copilot. - [Start with Grok](https://trywisp.dev/docs/guides/agents/grok.md): Run Grok Build (the grok CLI) in wisp on your own xAI account or key. Pick its model and effort, and open a task back in Grok. - [Start with Antigravity](https://trywisp.dev/docs/guides/agents/antigravity.md): Run the Antigravity CLI (agy) in wisp on your own Google account or Gemini key. Pick its model and effort, and open a task back in Antigravity. - [The desktop app](https://trywisp.dev/docs/guides/apps/desktop-app.md): wisp as an app on macOS, Windows and Linux. Install it, open it, keep it up to date, and what the window adds over a browser tab. - [The phone app](https://trywisp.dev/docs/guides/apps/phone-app.md): The wisp apps for iOS and Android are coming. Join the waitlist, and use wisp from your phone's browser and notifications until then. - [The terminal UI](https://trywisp.dev/docs/guides/apps/tui.md): wisp full screen in your terminal with `wisp tui`. Its zones, every key, the palette, `/` commands and the views. - [Build a software factory as code](https://trywisp.dev/docs/guides/declare/software-factory-as-code.md): Declare agents that pick up issues, build, review and keep the repo healthy, in your repo's .wisp/ folder. - [Share skills, AGENTS.md and MCP servers with your team](https://trywisp.dev/docs/guides/declare/share-skills-and-mcps.md): Give every agent on your team the same rules, skills and MCP servers, on every harness, in one repo or across many. - [Run a task under a formula, and reference everything by name](https://trywisp.dev/docs/guides/declare/run-under-a-formula.md): Try a setup on one task without applying it, run a task a formula declares by its name, and keep task files that point at your formula instead of copying it. - [Run a task and follow up](https://trywisp.dev/docs/guides/run/run-and-follow-up.md): Start a task, watch it, follow up, queue, ask on the side, stop it, and start a turn that wisp held. - [Compare agents and try it two ways](https://trywisp.dev/docs/guides/run/compare-and-try.md): Run one task on several harnesses side by side, or twice on one harness with one setting changed, then keep the better run. - [Run a task through a software factory](https://trywisp.dev/docs/guides/run/software-factory.md): Spec you approve, build, checks, review and a PR you merge, with failures sent back to build, from the app, the phone, the TUI or the CLI. - [Worktrees and shipping a PR](https://trywisp.dev/docs/guides/run/worktrees-and-shipping.md): Each task works on its own branch in its own git worktree. Read its changes, ship a pull request, watch its checks, fix conflicts and merge. - [Goals and judges](https://trywisp.dev/docs/guides/run/goals.md): Give a task a goal and judges. wisp sends the agent back after each turn until the judges agree the goal is met. - [Review by another agent](https://trywisp.dev/docs/guides/run/review-by-another-agent.md): Have a second harness read a task's pull request before you merge, ask a crew reviewer, or delegate a review. - [Crew and check-ins](https://trywisp.dev/docs/guides/run/crew-and-check-ins.md): Add crew members, give them jobs, let them check in on their own, set who hands work to whom, and follow them as stories on your phone. - [Automations, schedules and triggers](https://trywisp.dev/docs/guides/run/schedules-and-triggers.md): Save prompts as recipes, start tasks on a schedule or when something happens (CI, issues, PRs, chat, webhooks), and follow up later. - [Call wisp and your crew from Slack, Linear and more](https://trywisp.dev/docs/guides/run/from-slack-and-linear.md): Start tasks, talk to crew members and answer questions from Slack, Discord, Telegram, email, GitHub, Linear, ntfy and your phone. - [Long threads and fresh threads](https://trywisp.dev/docs/guides/run/long-threads.md): Keep a long task on track with its Notes, see how worn its thread is, and start a fresh conversation on the same task. - [Machines, the Grid and wisp Cloud](https://trywisp.dev/docs/guides/run/machines.md): Run tasks on your other computers over SSH or a wisp Grid, get a cloud computer on wisp Cloud, run your own Grid, and give a task a sandbox of its own. - [Budgets and usage](https://trywisp.dev/docs/guides/run/budgets-and-usage.md): See how much of your plans' usage windows your tasks take, have wisp ask before a task goes past your line, and keep automations from using up a window. - [MCP from your harness](https://trywisp.dev/docs/guides/run/mcp-from-your-harness.md): Give Claude Code, Codex, opencode or any MCP client wisp's tools with `wisp mcp`, so your agent can start wisp tasks, wait for them and follow up. - [Ask about what just happened in your terminal](https://trywisp.dev/docs/guides/run/terminal-context.md): A command failed. Type wisp "check the error above" and the agent gets the command, its exit code, its output and your folder. - [Track your agents' work beside your code](https://trywisp.dev/docs/guides/track/track-agent-work.md): Keep every task that wrote your code in the repo, find the task behind any line, and pick up where it left off. ## Integrations - [Use wisp from your agent or editor](https://trywisp.dev/docs/integrations/your-agent.md): Give Claude Code, Codex or any MCP client wisp's tools, so it can start tasks, read history and hand off work. ## Build plugins - [Make wisp your own](https://trywisp.dev/docs/plugins/make-wisp-your-own.md): Change how wisp works, from a setting to a plugin an agent writes for you, with the same API wisp's own parts use. - [Your first plugin](https://trywisp.dev/docs/plugins/first-plugin.md): Write a plugin in any language that adds a command to wisp, then call it from the CLI, the app, the SDKs and MCP. ## Reference - [CLI](https://trywisp.dev/docs/reference/cli.md): Every wisp command and option, as `wisp --help` prints it. - [Commands](https://trywisp.dev/docs/reference/commands.md): Every command each built-in plugin contributes, from its manifest. - [Settings](https://trywisp.dev/docs/reference/settings.md): Every setting each built-in plugin declares, with its type and default. - [MCP tools](https://trywisp.dev/docs/reference/mcp-tools.md): The tools wisp gives an agent running in a task, and the ones `wisp mcp` gives your own agent or editor. - [Answer blocks](https://trywisp.dev/docs/reference/answer-blocks.md): The blocks an agent ends its answer with, which wisp draws: the shape and the rule for each. - [Resource kinds](https://trywisp.dev/docs/reference/kinds.md): Every kind a formula can declare, with its fields, from the schema wisp checks it with. - [Plugin manifest and UI](https://trywisp.dev/docs/reference/plugin-api.md): Every key of a plugin's wisp-plugin.json, and the components a plugin's UI can use, from the checker's own data. - [Events and trigger sources](https://trywisp.dev/docs/reference/triggers.md): Every value a trigger's `on` takes, and the plugin that offers it. - [Plugin examples](https://trywisp.dev/docs/reference/plugin-examples.md): Whole plugins, manifest and code, that wisp checks in its own tests: copy one into ~/.wisp/plugins/ to start. ## Import a wisp home - [Import a wisp home](https://trywisp.dev/docs/migrate/import.md): Bring projects, recipes, crew, automations, machines and task history from another wisp home (~/.wisp) with one command.