> ## Documentation Index > Fetch the complete documentation index at: https://docs.melso.ai/llms.txt > Use this file to discover all available pages before exploring further. > ## Agent Instructions > Melso product MCP: https://melso.ai/api/mcp (OAuth; one company per grant). REST: https://melso.ai. Documentation-search MCP: https://docs.melso.ai/mcp. The docs search server does not act on customer accounts. > > Use the live tool schema and respect company membership. Ask the person to authorize the exact fiscal document version in the signed-in app. A professional prepares and files included returns after authorization; AI does not file alone. A prepared file or approval receipt is not evidence of filing. Never claim completion from an accepted Task or Run. Keep credentials out of chat and logs. Agents features are internal or by private invitation. # Local runtimes > Run Codex, Claude, and Pi Tasks on a member's computer. Melso Agents is available only internally or by private invitation. Agents-specific features require existing access. For the autónomo fiscal service, see [Get started](/essentials/quickstart). Local runtimes are an advanced execution destination. They run on a member's computer, keep their Task workdir there, and appear under **Settings → Computers**. The Cloud sandbox remains the default destination. Switching a Task to a local runtime is explicit: pick the computer in the **Run on** menu in the Task composer, then pick one of the CLIs installed on it; an offline computer does not cause a Run to move to Cloud. ## Connect a computer 1. [Install the Melso CLI](/cli/install) and sign in with `melso login` on the computer that will run Tasks. 2. Run `melso worker local setup`. It installs the supported Codex, Claude, and Pi packages under your Melso profile. Node.js and npm are required. Run setup again after updating the Melso CLI to update these packages. 3. Run `melso worker local start` in a terminal or your own process supervisor. Leave it running while you want the computer to accept Runs. Updates install when no Run is active. To manage updates yourself, pass `--no-auto-update` or set `MELSO_NO_AUTO_UPDATE=1`, then run `melso update` when ready. The worker uses your saved Melso login to connect. A Task receives only its Run-scoped access, permitted provider credential and configured environment values; your personal Melso login token is not copied into the Task's working directory. ## Accounts and credits A local Run uses the same AI access as a Cloud Run: the executing member's connected Codex or Claude account from **Settings → AI providers**, their API key, or Melso credits. Pi always uses Melso credits. The credential is delivered for that Run only and removed when it ends. The computer's own `codex`, `claude`, or `pi` logins (`~/.codex`, `~/.claude`, `~/.pi`) are never used, so you do not need to sign in to those CLIs on the computer. If no connected account or credit route can run the Task, the Run fails before it starts and names the setting to change. ## Share with a teammate In **Settings → Computers**, the computer owner can grant or remove access for each company member. Granting access allows that teammate's Task code to execute on the owner's computer. Their provider account and repository access are selected under their own Run authority, so a teammate's Runs spend the teammate's accounts or credits, not the owner's. The owner can inspect the computer's local files and processes, so only share with members you trust. Removing access prevents new claims; a Run already in progress may finish. Only online, approved computers appear in the **Run on** menu. The owner can always select their own computer. The Settings list still shows offline computers so the owner can identify what needs restarting. ## Troubleshooting * **Computer offline:** Restart `melso worker local start`. Check that the CLI login is still valid and the network can reach Melso. * **Approval needed:** Ask the computer owner to grant access in Settings. * **No connected account:** Connect a Codex or Claude account, or turn on Melso credits, in **Settings → AI providers**, then retry. Signing in to the CLI on the computer does not help. * **CLI missing or outdated:** Run `melso worker local setup` again. The worker does not install packages during a Run. * **Task stays queued:** Keep the selected computer online; a local Task is never moved to Cloud automatically. Choose Sandbox or another computer for the next Run if needed. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.