agentleFS
millhqmillhq/mill/userdocs/llms.txt

mill / userdocs

llms.txt1 starsChanged 2 months agoApache-2.0View on GitHub

Mill is a desktop app for guardrailed automation: workflows composed from typed steps, run by hotkey/schedule/watcher, with every external effect gated for approval — and a full MCP surface so AI agents can drive it under the same guardrails. Single binary, local data, no phone-home.

The file

Files
llms.txt
# Mill

> Mill is a desktop app for guardrailed automation: workflows composed from typed steps, run by hotkey/schedule/watcher, with every external effect gated for approval — and a full MCP surface so AI agents can drive it under the same guardrails. Single binary, local data, no phone-home.

## Start here

- [What is Mill](start-here/what-is-mill.md): the product in three ideas
- [Install](start-here/install.md): release install, source build, where data lives
- [Your first workflow](start-here/first-workflow.md): run the seeded example, then rebuild it
- [Your first board](start-here/first-board.md): place a card, a table and a diagram on a board, line them up, undo

## How-to

- [Store and reference a secret](how-to/store-and-reference-a-secret.md): put a value in the vault and pick it wherever a step needs it
- [Fire a workflow from a webhook](how-to/webhooks.md): minting a webhook token and posting to Mill's webhook address so a tool or service fires a workflow
- [Reply to a webhook](how-to/reply-to-a-webhook.md): the Answer the webhook step, reading the reply from a plain HTTP request or curl, and why step order decides what a run answers with
- [Install a plugin](reference/install-a-plugin.md): installing from a marketplace, a repository, or a folder, and what the badge means
- [Port an extension from another platform](reference/port-a-vscode-extension.md): a field-by-field mapping table, which parts need no rewrite, and which ones a compatible port always reauthors

## Concepts

- [Workflows and steps](concepts/workflows-and-steps.md): triggers, the typed step contract, payload vs attributes, versions
- [Guardrails and effect classes](concepts/guardrails.md): what asks for approval and how rules scope it
- [Configure entities](concepts/configure.md): integrations, lists, MCP servers, AI providers, environments
- [Selecting and bulk actions](concepts/selecting-and-bulk-actions.md): the checkbox, Shift/⌘-click, long-press, the selection bar, and bulk delete with undo, shared by every list
- [Atlas](concepts/atlas.md): the knowledge board: kinds, links, areas, doc mirrors, card actions
- [Runs, review, and debugging](concepts/runs-and-review.md): durable runs, the review queue, breakpoints
- [The coding loop](concepts/coding-loop.md): copy a command, confirm the parsed steps, watch it run, copy the result back
- [Secrets are references](concepts/secrets.md): the vault, references in every secret field, sources, the lock policy
- [Extensions and trust](concepts/extensions.md): the store, sources, verification tiers, and what an extension can reach
- [Trust, data, and safety](trust/data-and-safety.md): no phone-home, local data, honest limits

## Reference

- [Step reference](reference/steps.md): every step's contract, generated from the registry
- [Commands](reference/commands.md): every registered command's id, label, default binding, surface, and enablement, generated from the registry
- [Menu bar](reference/menu-bar.md): every menu and item in the macOS menu bar, its shortcut, and the command behind it, generated from the registry
- [Environments](reference/environments.md): named sets of plain and secret variables, {{var}} in a request, and which one a run selects
- [Client certificates](reference/client-certificates.md): presenting a client certificate per host: formats, matching, expiry status, and the Test action
- [The browser extension](reference/browser-extension.md): pairing a browser so Mill can replay recorded steps in your own signed-in session
- [Settings](reference/settings.md): app preferences: appearance, vault lock policy, hotkeys, shortcuts, MCP access, remote access, backups, updates
- [Extending the canvas](reference/extending-the-canvas.md): how a canvas noun loads, what its declaration requires, and what platform APIs it may and may not reach
- [Register a canvas tool](reference/register-a-canvas-tool.md): walks a new AtlasToolShape declaration end to end, quoting a real registered tool
- [Register a command](reference/register-a-command.md): walks a new Command registry entry end to end, quoting a real registered command
- [The plugin standard](reference/plugin-standard.md): the rules a shipped plugin follows, and which ones the conformance checker enforces
- [Plugin theming](reference/plugin-theming.md): the theme variables a plugin renders with
- [Managed extensions](reference/managed-extensions.md): the organisation policy file: allow and block lists, required tier, blocked capabilities, allowed sources, and what an install checks
- [Plugin API maturity](reference/plugin-api-maturity.md): each contribution family's level and its proof, generated from the repository

## For agents

- [Automate with agents](agents/connect-mcp.md): connecting over MCP and what agents can do
- [Edit a diagram with an agent](agents/diagrams.md): reading a diagram's shapes by id and adding, changing, deleting and importing them in place
- [Edit a sheet with an agent](agents/sheets.md): reading a sheet's cells by range and changing named cells in place
- [Edit a spreadsheet file with an agent](agents/spreadsheets.md): reading a spreadsheet file's cells and formulas by range and changing named cells in place, everything else preserved
- [Add a row with an agent](agents/lists.md): reading a List's columns and rows and appending a new row to an existing one
- [What plugins expose to agents](agents/plugins.md): listing installed plugins, calling a plugin's declared tools, and how a plugin write parks

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.