Human-facing React components built on Base UI and Tailwind CSS, with agent-readable contracts and optional tests.
Shows the UI an agent composed in A2UI, with Ouro HX: it resolves the messages into core's composed tree, draws it as ComposedSurface does, and sends a press as A2UI's action message.
Optional: the Vitest browser tests, type tests, stories and Playwright spec for a2ui-surface, and the tests of everything it installs. Add it beside a2ui-surface: add a2ui-surface a2ui-surface-tests. It names Vitest, Playwright and Storybook without a version, so the CLI installs one only when your app lacks it and never changes a version you have.
A stack of titles that each open their own content, for FAQs and long settings pages.
Optional: the Vitest browser tests, type tests, stories and Playwright spec for accordion, and the tests of everything it installs. Add it beside accordion: add accordion accordion-tests. It names Vitest, Playwright and Storybook without a version, so the CLI installs one only when your app lacks it and never changes a version you have.
The person's always-visible sign of which of their agents are listening to something on their behalf, such as a meeting, with Stop one click away.
Optional: the Vitest browser tests, type tests, stories and Playwright spec for agent-presence, and the tests of everything it installs. Add it beside agent-presence: add agent-presence agent-presence-tests. It names Vitest, Playwright and Storybook without a version, so the CLI installs one only when your app lacks it and never changes a version you have.
An inline message in the page: something saved, something failed, something to know before going on.
Optional: the Vitest browser tests, type tests, stories and Playwright spec for alert, and the tests of everything it installs. Add it beside alert: add alert alert-tests. It names Vitest, Playwright and Storybook without a version, so the CLI installs one only when your app lacks it and never changes a version you have.
A frame that keeps an image, video or map at a fixed shape while its width follows the layout, so the page does not jump when the media loads.
Optional: the Vitest browser tests, type tests, stories and Playwright spec for aspect-ratio, and the tests of everything it installs. Add it beside aspect-ratio: add aspect-ratio aspect-ratio-tests. It names Vitest, Playwright and Storybook without a version, so the CLI installs one only when your app lacks it and never changes a version you have.
A file attached to a chat message, or waiting to be sent with one: an icon for its kind, its name and size.
Optional: the Vitest browser tests, type tests, stories and Playwright spec for attachment, and the tests of everything it installs. Add it beside attachment: add attachment attachment-tests. It names Vitest, Playwright and Storybook without a version, so the CLI installs one only when your app lacks it and never changes a version you have.
Something that waits on the person: a mention, a question asked of them or a request still unanswered.
Optional: the Vitest browser tests, type tests, stories and Playwright spec for attention-card, and the tests of everything it installs. Add it beside attention-card: add attention-card attention-card-tests. It names Vitest, Playwright and Storybook without a version, so the CLI installs one only when your app lacks it and never changes a version you have.
The page for signing in, signing up or resetting a password: one `Card` in the middle of the screen, with your brand, the title as the page's `<h1>`, the form, and a line under the card.
Optional: the Vitest browser tests, type tests, stories and Playwright spec for auth-layout, and the tests of everything it installs. Add it beside auth-layout: add auth-layout auth-layout-tests. It names Vitest, Playwright and Storybook without a version, so the CLI installs one only when your app lacks it and never changes a version you have.
Suggestions at the text caret: typing a trigger character, such as `@` for people or `:` for emoji, in a `TextArea` or `TextField` opens a list beside the caret, filtered by what follows the character.
Optional: the Vitest browser tests, type tests, stories and Playwright spec for autocomplete, and the tests of everything it installs. Add it beside autocomplete: add autocomplete autocomplete-tests. It names Vitest, Playwright and Storybook without a version, so the CLI installs one only when your app lacks it and never changes a version you have.
A person's picture, or their initials until it loads or when it fails, in a color picked from the name.
Optional: the Vitest browser tests, type tests, stories and Playwright spec for avatar, and the tests of everything it installs. Add it beside avatar: add avatar avatar-tests. It names Vitest, Playwright and Storybook without a version, so the CLI installs one only when your app lacks it and never changes a version you have.
Ouro AX: A tool's checks, such as a doctor or a self-check, declared once: progress while they run, the verdict first and the results grouped for a person, the same facts as `--json` for an agent, the time taken and the exit code.
Optional: the node:test tests for ax-checks, in scripts/ouro/ax/; run them with node --test scripts/ouro/ax/*.test.mts.
Ouro AX: the Ouro AX standard as data: a claim is one checkable rule about an interface an agent uses, with the reason, the evidence that admitted it, the check `ax-xray` runs and the fix.
Optional: the node:test tests for ax-claims, in scripts/ouro/ax/; run them with node --test scripts/ouro/ax/*.test.mts.
Ouro AX: the whole front door for a tool built on the kit: parse, help, run (or, on `--dry-run`, plan), print the result as text or `--json`, and turn every failure into the one error line and its exit code.
Optional: the node:test tests for ax-cli, in scripts/ouro/ax/; run them with node --test scripts/ouro/ax/*.test.mts.
Ouro AX: one definition of a command: its words, effect, arguments and flags, examples and output.
Optional: the node:test tests for ax-command, in scripts/ouro/ax/; run them with node --test scripts/ouro/ax/*.test.mts.
Ouro AX: A component's agent side: what an agent may read from it and which of its actions an agent may take, each bound to the CLI command that does the same (its twin).
Optional: the node:test tests for ax-component, in scripts/ouro/ax/; run them with node --test scripts/ouro/ax/*.test.mts.
Ouro AX: A message for an agent or a person names only what the installed tool runs: never a developer command, a repository path or a step that only works in a source checkout.
Optional: the node:test tests for ax-dev-commands, in scripts/ouro/ax/; run them with node --test scripts/ouro/ax/*.test.mts.
Ouro AX: effect classes: a command either touches only the person's own space ("private") or reaches other people ("others").
Optional: the node:test tests for ax-effects, in scripts/ouro/ax/; run them with node --test scripts/ouro/ax/*.test.mts.
Ouro AX: a small tool built only from the kit, the pattern to copy; the ax tests use it.
Ouro AX: A tool's exit codes as one table: the overview, every command's footer, `help --json` and the MCP error shape read it, so the numbers never drift.
Optional: the node:test tests for ax-exit-codes, in scripts/ouro/ax/; run them with node --test scripts/ouro/ax/*.test.mts.
Ouro AX: an adopter's agent flows as data, run in CI without a model: a flow is the calls a good agent makes for one task, and the runner checks the calls' exit codes, JSON shapes and fields, error text, time, size and voice.
Optional: the node:test tests for ax-flows, in scripts/ouro/ax/; run them with node --test scripts/ouro/ax/*.test.mts.
Ouro AX: the kit's check of a tool's declarations: what an agent needs from every declared command (a summary, an example, described arguments and flags, `--why` and a preview where other people see it, help that fits the width), and, against the tool's own command inventory, that every command it registers runs through its declaration.
Optional: the node:test tests for ax-lint, in scripts/ouro/ax/; run them with node --test scripts/ouro/ax/*.test.mts.
Ouro AX: the same definitions as an MCP server over stdio (newline-delimited JSON-RPC 2.0), with no dependencies: one tool per command, its inputSchema from the definition, calls checked before they run with the fix in the one error shape, effects as tool annotations, and results cut to a size an agent's context can hold.
Optional: the node:test tests for ax-mcp, in scripts/ouro/ax/; run them with node --test scripts/ouro/ax/*.test.mts.
Ouro AX: the outcome of an action: what became of it, who can see it, whether that was checked, how to find it again and what to run next, in one shape over a CLI's `--json`, an HTTP route and an MCP tool.
Optional: the node:test tests for ax-outcome, in scripts/ouro/ax/; run them with node --test scripts/ouro/ax/*.test.mts.
Ouro AX: one result, two renderings: `--json` prints the data, text prints what the command's text() makes of the same data.
Optional: the node:test tests for ax-output, in scripts/ouro/ax/; run them with node --test scripts/ouro/ax/*.test.mts.
Ouro AX: A failure says what happened and precisely why (the system and its answer: "The mail service says this file was deleted"), who or what caused it where that is known, and the next step as a working action.