> ## Documentation Index > Fetch the complete documentation index at: https://ade-app.dev/docs/llms.txt > Use this file to discover all available pages before exploring further. # ade CLI > The ade command is the typed control plane for ADE from any shell — the same lanes, chats, PRs, and tests, scriptable for humans and agents alike. `ade` is the typed control plane for ADE. Everything you do in the desktop app — create lanes, run agents, ship PRs, kick off tests, attach Linear issues — has a CLI command, so you can drive ADE from any shell. Agents running inside ADE use the same commands, which is how they inspect and operate the workspace they live in. ```bash theme={null} ade lanes create "fix-checkout-flow" # spin up a worktree for a task ade brain status --text # check the machine runtime ``` > [!TIP] > > Two commands discover everything: `ade --help` for exact flags on any command, and `ade actions list --text` for the full service-backed action catalog behind the typed commands. ## Output modes Most commands take an explicit output mode. Use `--text` for a human-readable summary and `--json` for stable, scriptable fields (the default when output is piped). ```bash theme={null} ade lanes list --text # readable summary ade lanes list --json # stable JSON for scripts and agents ``` ## Command groups > [accordion] Project & lanes > > Register projects and manage worktrees (lanes). > > ```bash theme={null} > ade init # add the current directory as a project > ade projects list --text # list registered projects > ade lanes list --text # list lanes in the active project > ade lanes create "fix-checkout-flow" # create a lane from the default branch > ade lanes create --name auth-ui --from auth-base --on local > ade lanes delete --force --delete-branch > ``` > [accordion] Chats & sessions > > Start and inspect agent chats, terminals, and CLI sessions inside a lane. > > ```bash theme={null} > ade chat list --lane --text > ade chat create --lane --model gpt-5.5 > ade shell start-cli codex --lane --permission-mode edit --message "fix failing tests" > ade shell start --lane -- npm test > ``` > [accordion] Git & PRs > > Commit, push, and turn a lane into a pull request; read CI and comments. > > ```bash theme={null} > ade git commit --lane > ade git push --lane > ade prs create --lane --base main --title "Fix checkout flow" --text > ade prs list-open --text > ade prs checks --text # CI status for a PR > ade prs comments --text > ``` > [accordion] Tests > > Run test suites in a lane. > > ```bash theme={null} > ade tests run --lane --suite unit --wait > ``` > [accordion] Linear > > Attach and update Linear issues through ADE's own Linear connection — no API key needed. > > ```bash theme={null} > ade lanes create --name ENG-431 --issue-id ENG-431 --start-chat --provider codex --type subagent > ade linear attach --this-session --issue-id ENG-431 > ade linear comment "Pushed a fix; CI running" > ade linear set-state ENG-431 > ``` > [accordion] Account & machines > > Link this machine to your ADE account so you can reach it from ADE desktop, > the web client, and iOS. `ade connect` is idempotent — re-running it on a > connected machine just prints status. > > ```bash theme={null} > ade connect # sign in, install the service, publish this machine > ade connect --status --text # report the three steps, change nothing > ade connect --headless # sign in from another device: QR code, link, and code > ade connect --no-login # service only; stay local/LAN-only > ade connect --no-service # account only; skip the login service > ade connect --timeout 120 # seconds to wait for the machine row (default 60) > ade login # sign in without touching the service > ade machines list --text # machines registered to this account > ade machines reconnect # re-pair this machine after it was removed > ade logout # undo the account half of connect > ``` > > The machine row is published by the running brain, not by this command — so > the brain must stay running for the machine to remain reachable. Full > walkthrough, including running the brain on a Linux server: > [Connect a machine](https://ade-app.dev/docs/getting-started/connect-machine.md). > > Removing a machine from your account is deliberate and terminal: heartbeats > never re-register it, and its Activity, notifications, and Live Activities > stop. `ade machines reconnect` is the way back. It takes no selector — it > always re-pairs the machine you run it on — and it needs proof of a freshly > completed interactive sign-in, so it prompts for one when the account > directory asks and finishes the re-pair as soon as that lands. Only once the > directory accepts does it lift the push gate, so Activity, alerts, and > badges resume without restarting the brain. A normal `ade login` runs the > same repair automatically, and ADE desktop offers it as **Reconnect this > computer** in Settings. > [accordion] Brain & pairing > > Control the machine-owned ADE runtime (the brain) and pair a phone. > > ```bash theme={null} > ade brain status --text # endpoint, service, and sync state > ade brain start # load the login service > ade brain stop # unload the login service > ade brain restart # re-exec after an app update > ade brain pin generate # generate a phone pairing PIN > ade brain pin set 123456 # set an explicit pairing PIN > ``` > [accordion] Proof > > List the computer-use proof artifacts ADE captured for a chat or session. > > ```bash theme={null} > ade proof list --arg ownerKind=chat --arg ownerId= > ``` > [accordion] Discovery > > Find commands, flags, and the full action catalog, and check local readiness. > > ```bash theme={null} > ade --help # exact flags for any command > ade actions list --text # full service-backed action catalog > ade actions run git.stageFile --arg laneId= --arg path=src/index.ts > ade doctor # local runtime, git, provider readiness > ade triage # hand a broken install to a coding agent ADE has downloaded > ade triage --agent --text # print the context + playbook paths instead of launching one > ade skill list --text # bundled agent skills > ``` > [!NOTE] > > Prefer typed commands — they validate arguments and return stable fields. Reach for `ade actions run ` only when no typed command covers the workflow yet. ## Full reference For every command, subcommand, flag, and the runtime modes behind them, see the complete CLI README on GitHub. > [card] [ade CLI README](https://github.com/arul28/ADE/blob/main/apps/ade-cli/README.md) > > The full `ade` command reference, runtime modes, and install paths. > [card] [Architecture](https://ade-app.dev/docs/architecture.md) > > How the control plane, runtime, and clients fit together. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.