# H Platform Docs

> Documentation for H's computer-use platform: the Agents API, the Holo Models API, HoloDesktop CLI, and the HoloTab Chrome extension.

> ## Agent Instructions
> H Platform has four products: the Agents API (managed computer-use agents, base URL https://agp.eu.hcompany.ai/api/v2 or https://agp.hcompany.ai/api/v2 for the US), the Models API (OpenAI-compatible Holo vision-language models at https://api.hcompany.ai/v1), HoloDesktop CLI (Holo on the user's own desktop), and HoloTab (a free no-code Chrome extension that runs Holo in the user's browser, with recordable routines and schedules).
> Authenticate with a bearer API key from the HAI_API_KEY environment variable. SDKs: `pip install hai-agents` (Python, `from hai_agents import Client`) and `npm install hai-agents` (TypeScript, `import { HaiAgentsClient } from "hai-agents"`). CLI: `hai`.
> Agents do work in a browser or on a desktop; describe the task as an imperative instruction. To run a task quickly, prefer the pre-built agent `h/web-surfer-flash`. Read results from the session's `latest_answer` after it reaches a terminal status.
> Sessions are the unit of work; wait for a terminal status (completed, failed, timed_out, interrupted) before reading the answer. Use webhooks or the `changes` long-poll endpoint to follow progress.

- [H Platform Docs](https://hub.hcompany.ai/index.md): Run computer-use agents in a cloud browser or on a desktop with the Agents API, build your own agent on Holo models, or run Holo on your own machine.
- [Agents API](https://hub.hcompany.ai/agents-api/introduction.md): Run Holo-powered computer-use agents that do work in a cloud browser or on a desktop: pre-built agents, custom agents, live sessions, skills, vaults.
- [Agents API quickstart](https://hub.hcompany.ai/agents-api/quickstart.md): Get an API key, install the hai-agents SDK or CLI, run a pre-built computer-use agent with one call, then customize the environment and agent.
- [Run from the platform](https://hub.hcompany.ai/agents-api/run-from-the-platform.md): Run a pre-built computer-use agent from the H Platform without code: pick an agent, describe the task, start the session, watch it live, read the result.
- [Agents](https://hub.hcompany.ai/agents-api/agents/overview.md): A reusable configuration: environment, model, skills, instructions. Run a pre-built h/ agent or create your own with the Agents API.
- [Environments](https://hub.hcompany.ai/agents-api/environments/overview.md): Environments are what an agent sees and acts on: a cloud browser, your own browser, or your own desktop. Pre-built environments and how to define your own.
- [Sessions](https://hub.hcompany.ai/agents-api/sessions/overview.md): A session is one run of an agent on a task: lifecycle states, the session object, overrides, ways to wait for the result, listing and filtering.
- [Skills](https://hub.hcompany.ai/agents-api/skills/overview.md): Skills are reusable instruction fragments an agent draws on when a task matches: built-in h/ skills, writing your own, and attaching them to agents.
- [Vaults](https://hub.hcompany.ai/agents-api/vaults/overview.md): Let an agent sign in to the sites and services it works on, without passing secrets through the API.
- [Configure the browser](https://hub.hcompany.ai/agents-api/browser/configuration.md): Define a Browser environment: its fields, modes, and actions.
- [Bring your own browser profile](https://hub.hcompany.ai/agents-api/browser/profiles.md): Start a cloud browser session from saved cookies and storage: build, upload, persist, and default browser profiles.
- [Route the browser through a proxy](https://hub.hcompany.ai/agents-api/browser/proxy.md): Egress the browser through an H-managed proxy or one you operate.
- [Run the agent in your own browser](https://hub.hcompany.ai/agents-api/browser/local-control.md): Let a computer-use agent drive Chrome on your own machine through the local runtime, with your logins and extensions.
- [Run the agent on your own desktop](https://hub.hcompany.ai/agents-api/desktop/local-control.md): Let an agent drive the mouse, keyboard, and screen on your own machine.
- [Build with the SDKs](https://hub.hcompany.ai/agents-api/sdks.md): Install and configure the hai-agents Python and TypeScript SDKs and CLI: authentication, region, retries, async client, and example recipes.
- [Run agents over MCP](https://hub.hcompany.ai/agents-api/mcp.md): Run and manage H computer-use agents from Claude, Cursor, or any MCP host, with no HTTP or SDK code.
- [Build with coding assistants](https://hub.hcompany.ai/agents-api/coding-skills.md): Install the hai-agents skill so Claude Code, Cursor, and other assistants know the H APIs and help you build use cases.
- [Run agents from n8n](https://hub.hcompany.ai/agents-api/n8n.md): Run computer-use agents from an n8n workflow over plain HTTP: create the session, wait for the result, use the answer downstream.
- [Parallelize work with subagents](https://hub.hcompany.ai/agents-api/multi-agent.md): Let one manager agent split a task across specialist subagents, run them in parallel as child sessions, and merge their results.
- [Watch and steer sessions](https://hub.hcompany.ai/agents-api/observe-and-steer.md): Watch a run, intervene while it works, keep the conversation going, and read how it ended.
- [Give an agent custom tools](https://hub.hcompany.ai/agents-api/custom-tools.md): Extend the agent's toolbox with functions from your own code: the agent calls them, the SDK executes them, the run continues with the result.
- [Handle two-factor authentication](https://hub.hcompany.ai/agents-api/two-factor-auth.md): Let the agent request one-time codes and confirmation links through a built-in tool.
- [Get typed answers](https://hub.hcompany.ai/agents-api/structured-output.md): Get the agent's final answer as typed, schema-validated JSON: set answer_format on the agent or pass a Pydantic or Zod schema in the SDK.
- [Get notified with webhooks](https://hub.hcompany.ai/agents-api/webhooks/overview.md): Receive signed HTTP notifications when sessions change status, verify signatures, and rotate secrets.
- [Run sessions on a schedule](https://hub.hcompany.ai/agents-api/schedules/overview.md): Run a session on a recurring cron cadence, with pause, resume, manual trigger, and run history.
- [Plans and limits](https://hub.hcompany.ai/agents-api/plans-and-limits.md): Plans, token and concurrency allowances, and how to manage your subscription.
- [Handle API errors](https://hub.hcompany.ai/agents-api/errors.md): Agents API error envelope, HTTP status codes from 400 to 504 with causes and fixes, and how the SDKs raise and retry errors.
- [Changelog](https://hub.hcompany.ai/agents-api/changelog.md): Notable changes to the Agents API, the hai-agents SDKs, and the hai CLI, by release.
- [API reference](https://hub.hcompany.ai/agents-api/api-reference.md): Agents API base URLs per region, bearer authentication, JSON conventions, error envelope, pagination, idempotency, limits, and resources.
- [Create an agent](https://hub.hcompany.ai/agents-api/agents/create.md): Create a reusable agent in your catalog: model, environments, skills, instructions, subagents, and answer format, referenced by name in every session.
- [List agents](https://hub.hcompany.ai/agents-api/agents/list.md): Discover available agents from your catalog and the H preset catalog.
- [Retrieve an agent](https://hub.hcompany.ai/agents-api/agents/retrieve.md): Get the full specification of a single agent by name, including pre-built h/ agents.
- [Update an agent](https://hub.hcompany.ai/agents-api/agents/update.md): Replace the whole configuration of an agent in your catalog; later sessions pick up the new spec.
- [Patch an agent](https://hub.hcompany.ai/agents-api/agents/patch.md): Change individual fields of an agent without resending the full object.
- [Delete an agent](https://hub.hcompany.ai/agents-api/agents/delete.md): Remove an agent from your catalog by name.
- [Create an environment](https://hub.hcompany.ai/agents-api/environments/create.md): Create a reusable environment in your catalog: a cloud browser, your own browser, or your own desktop, with its settings.
- [List environments](https://hub.hcompany.ai/agents-api/environments/list.md): List the environments available to your organization, including the pre-built h/ browsers.
- [Retrieve an environment](https://hub.hcompany.ai/agents-api/environments/retrieve.md): Get the full specification of an environment by id, including mode, start URL, vault, and profile settings.
- [Update an environment](https://hub.hcompany.ai/agents-api/environments/update.md): Replace the configuration of an environment in your catalog.
- [Patch an environment](https://hub.hcompany.ai/agents-api/environments/patch.md): Change individual fields of an environment without resending the full spec.
- [Delete an environment](https://hub.hcompany.ai/agents-api/environments/delete.md): Remove an environment from your catalog by id.
- [Create a session](https://hub.hcompany.ai/agents-api/sessions/create.md): Launch a session: run a pre-built or custom agent on a task, with optional overrides, idempotency key, and parent session.
- [List sessions](https://hub.hcompany.ai/agents-api/sessions/list.md): Browse your sessions with filtering, sorting, and pagination.
- [Retrieve a session](https://hub.hcompany.ai/agents-api/sessions/retrieve.md): Get the full session object: status, agent, environments, latest answer, timestamps, and usage.
- [Get session status](https://hub.hcompany.ai/agents-api/sessions/status.md): Lightweight polling endpoint: current state, step count, and token usage of a session.
- [Get session changes](https://hub.hcompany.ai/agents-api/sessions/changes.md): Long-poll a session for new events, status, and the final answer since an event index, with the polling rules and the SDK wait helper.
- [List events](https://hub.hcompany.ai/agents-api/sessions/events.md): Paginated list of everything the agent observed and did in a session, for replay and audit.
- [Send messages](https://hub.hcompany.ai/agents-api/sessions/send-messages.md): Send a message to a running agent to add context, correct course, or answer a question it asked.
- [Send tool results](https://hub.hcompany.ai/agents-api/sessions/tool-results.md): Answer the agent's pending custom tool calls so the session can continue.
- [Pause a session](https://hub.hcompany.ai/agents-api/sessions/pause.md): Halt a running session without losing its state, then resume it later.
- [Resume a session](https://hub.hcompany.ai/agents-api/sessions/resume.md): Continue a paused session from where the agent stopped.
- [Force an answer](https://hub.hcompany.ai/agents-api/sessions/force-answer.md): Ask a running agent to stop and emit its final answer on the next step.
- [Cancel a session](https://hub.hcompany.ai/agents-api/sessions/cancel.md): Stop a running session; the agent halts and the session ends as interrupted.
- [Get quota](https://hub.hcompany.ai/agents-api/sessions/quota.md): Check how many concurrent sessions your organization can run and how many are in use.
- [Session feedback](https://hub.hcompany.ai/agents-api/sessions/feedback.md): Report whether a session completed its task, to track agent quality over time.
- [Event feedback](https://hub.hcompany.ai/agents-api/sessions/event-feedback.md): Submit feedback on one event within a session to flag a wrong or good step.
- [Share a session](https://hub.hcompany.ai/agents-api/sessions/share.md): Make a session publicly viewable in Agent View through a share URL, for people outside your organization.
- [Unshare a session](https://hub.hcompany.ai/agents-api/sessions/unshare.md): Revoke a session's public share link so the Agent View URL stops working for outsiders.
- [Get resource](https://hub.hcompany.ai/agents-api/sessions/resources.md): Download session-owned resources such as screenshots and files the agent produced.
- [List files](https://hub.hcompany.ai/agents-api/sessions/list-files.md): List the files on a cloud browser's machine, such as what the agent downloaded.
- [Read file](https://hub.hcompany.ai/agents-api/sessions/read-file.md): Download a file from a cloud browser's machine, base64-encoded.
- [Write file](https://hub.hcompany.ai/agents-api/sessions/write-file.md): Upload a file to a cloud browser's machine for the agent to use.
- [Create a skill](https://hub.hcompany.ai/agents-api/skills/create.md): Create a reusable skill in your catalog: a description that routes it and a body the agent follows when it matches.
- [List skills](https://hub.hcompany.ai/agents-api/skills/list.md): Browse your skill catalog, including the built-in h/ skills, with pagination and name filtering.
- [Retrieve a skill](https://hub.hcompany.ai/agents-api/skills/retrieve.md): Get a skill's full details by name: description, body, source, and URL pattern.
- [Update a skill](https://hub.hcompany.ai/agents-api/skills/update.md): Edit a skill's description or body; agents pick up the change on their next session.
- [Patch a skill](https://hub.hcompany.ai/agents-api/skills/patch.md): Change individual fields of a skill without resending the full object.
- [Delete a skill](https://hub.hcompany.ai/agents-api/skills/delete.md): Remove a skill from your catalog by name.
- [Create a vault](https://hub.hcompany.ai/agents-api/vaults/create.md): Register a secrets provider so agents can sign in to sites without secrets passing through the API.
- [List vaults](https://hub.hcompany.ai/agents-api/vaults/list.md): Browse your organization's vault configs with offset pagination.
- [Retrieve a vault](https://hub.hcompany.ai/agents-api/vaults/retrieve.md): Fetch a vault config by id: provider, endpoint, and health, without the stored token.
- [Update a vault](https://hub.hcompany.ai/agents-api/vaults/update.md): Change a vault config's name or provider settings; the stored token stays as is.
- [Rotate the token](https://hub.hcompany.ai/agents-api/vaults/rotate-token.md): Replace a vault's stored service account token without changing its id or provider settings.
- [Delete a vault](https://hub.hcompany.ai/agents-api/vaults/delete.md): Remove a vault config by id.
- [Check health](https://hub.hcompany.ai/agents-api/vaults/health.md): Probe a vault's provider connection to confirm the token and endpoint work before a session needs them.
- [List browser profiles](https://hub.hcompany.ai/agents-api/browser-profiles/list.md): Browse your organization's browser profiles with offset pagination.
- [Initiate a profile upload](https://hub.hcompany.ai/agents-api/browser-profiles/initiate-upload.md): Get a presigned URL to upload a browser profile archive directly to storage.
- [Complete a profile upload](https://hub.hcompany.ai/agents-api/browser-profiles/complete-upload.md): Finalize a browser profile after its archive is uploaded to storage so sessions can load it.
- [Retrieve a browser profile](https://hub.hcompany.ai/agents-api/browser-profiles/retrieve.md): Fetch a browser profile by id, with its upload state and default flag.
- [Delete a browser profile](https://hub.hcompany.ai/agents-api/browser-profiles/delete.md): Remove a browser profile and its stored cookies and storage.
- [Set the default browser profile](https://hub.hcompany.ai/agents-api/browser-profiles/set-default.md): Mark a browser profile as the default that sessions load when none is named.
- [Get the default browser profile](https://hub.hcompany.ai/agents-api/browser-profiles/get-default.md): Fetch the browser profile sessions load by default when none is named.
- [Unset the default browser profile](https://hub.hcompany.ai/agents-api/browser-profiles/unset-default.md): Clear the default flag on a browser profile so sessions start from an empty browser.
- [Create a webhook](https://hub.hcompany.ai/agents-api/webhooks/create.md): Register a URL to receive signed notifications when sessions change status; the signing secret is returned once.
- [List webhooks](https://hub.hcompany.ai/agents-api/webhooks/list.md): Browse your organization's webhooks with their URLs, subscribed events, and delivery stats.
- [Retrieve a webhook](https://hub.hcompany.ai/agents-api/webhooks/retrieve.md): Fetch a single webhook by id, with its subscribed events and delivery stats.
- [Update a webhook](https://hub.hcompany.ai/agents-api/webhooks/update.md): Change a webhook's URL, events, description, or disabled state.
- [Delete a webhook](https://hub.hcompany.ai/agents-api/webhooks/delete.md): Remove a webhook so its URL stops receiving notifications.
- [List event types](https://hub.hcompany.ai/agents-api/webhooks/events.md): List the webhook event types you can subscribe to, with the payload each one carries.
- [Ping a webhook](https://hub.hcompany.ai/agents-api/webhooks/ping.md): Send a signed test event and see how your endpoint responds.
- [Rotate a webhook secret](https://hub.hcompany.ai/agents-api/webhooks/rotate.md): Replace a webhook's signing secret while both old and new secrets verify, so no delivery fails.
- [Create a schedule](https://hub.hcompany.ai/agents-api/schedules/create.md): Register a cron schedule that starts a session on each fire.
- [List schedules](https://hub.hcompany.ai/agents-api/schedules/list.md): Browse your organization's schedules with their cron cadence and status.
- [Retrieve a schedule](https://hub.hcompany.ai/agents-api/schedules/retrieve.md): Fetch a schedule by id, with its cron expression, status, and session template.
- [Update a schedule](https://hub.hcompany.ai/agents-api/schedules/update.md): Change a schedule's name, cron timing, or the session template it creates sessions from.
- [Delete a schedule](https://hub.hcompany.ai/agents-api/schedules/delete.md): Remove a schedule and stop future fires. Sessions it already created are unaffected.
- [Pause a schedule](https://hub.hcompany.ai/agents-api/schedules/pause.md): Stop a schedule's future fires without deleting it; resume it later with the same template.
- [Resume a schedule](https://hub.hcompany.ai/agents-api/schedules/resume.md): Restart a paused schedule so it fires again on its cron cadence.
- [Trigger a schedule](https://hub.hcompany.ai/agents-api/schedules/trigger.md): Fire a schedule once, immediately, without changing its cron cadence.
- [List schedule runs](https://hub.hcompany.ai/agents-api/schedules/runs.md): Browse a schedule's fire history: which sessions each fire created and how they ended.
- [Models API](https://hub.hcompany.ai/models-api/introduction.md): Holo models power agents that work through any interface: they click on screens, run code, and call your APIs and MCP tools. Which model to pick, open weights, pricing.
- [Models API quickstart](https://hub.hcompany.ai/models-api/quickstart.md): Get a Models API key, point the OpenAI client at https://api.hcompany.ai/v1, and make your first Holo call in Python, TypeScript, or cURL.
- [Core concepts](https://hub.hcompany.ai/models-api/build-an-agent/core-concepts.md): The harness conventions Holo is trained on. Get them right and either output format works.
- [Function calling](https://hub.hcompany.ai/models-api/build-an-agent/function-calling.md): The Holo agent loop with OpenAI-style tools and tool_calls, and the mistakes specific to this format.
- [Structured outputs](https://hub.hcompany.ai/models-api/build-an-agent/structured-outputs.md): The Holo agent loop with one constrained JSON object per step, and the mistakes specific to this format.
- [Browser agent](https://hub.hcompany.ai/models-api/cookbooks/browser-agent.md): A complete Holo browser agent in one Python file: Playwright takes screenshots, Holo returns clicks and keystrokes, the loop runs until the answer tool is called.
- [Code agent](https://hub.hcompany.ai/models-api/cookbooks/code-agent.md): A Holo agent with no screen: it writes Python, runs it, reads the output, and answers. One file, one tool.
- [MCP agent](https://hub.hcompany.ai/models-api/cookbooks/mcp-agent.md): Connect Holo to any MCP server: list its tools, pass them as OpenAI functions, and route each tool call back through the MCP session.
- [Hybrid agent](https://hub.hcompany.ai/models-api/cookbooks/hybrid-agent.md): One Holo agent with both a browser and a Python runner. It reads the page with clicks, then switches to code for the part a script does better.
- [Element localization](https://hub.hcompany.ai/models-api/element-localization.md): GUI grounding with Holo: send a screenshot and an element description, get click coordinates back.
- [Document OCR](https://hub.hcompany.ai/models-api/document-ocr.md): Turn a digitally rendered page into Markdown in one chat completion: headings, tables, equations, reading order.
- [Local inference](https://hub.hcompany.ai/models-api/local-inference.md): Holo4 and Holo3.1 are open weights. Serve them with llama.cpp or vLLM, point base_url at your server, and every example in these docs runs as is. Prompts, screenshots, and tool results never leave your network.
- [API reference](https://hub.hcompany.ai/models-api/api-reference.md): Models API base URL, bearer authentication, OpenAI-compatible endpoints, response channels, rate limits per tier, and error codes.
- [Create chat completion](https://hub.hcompany.ai/models-api/chat-completions.md): OpenAI-compatible chat completion with Holo-specific structured outputs and reasoning.
- [List models](https://hub.hcompany.ai/models-api/list-models.md): Programmatic discovery of the served models, their capabilities, pricing, and deprecation dates.
- [HoloTab](https://hub.hcompany.ai/holotab/introduction.md): HoloTab is a free Chrome extension that runs Holo in your browser. Describe a task in the side panel, watch it navigate and act, then record routines and put them on a schedule.
- [HoloTab quickstart](https://hub.hcompany.ai/holotab/quickstart.md): Install HoloTab from the Chrome Web Store, accept the safety notice, sign in, run a first task from the side panel, and write prompts that work.
- [Routines and schedules](https://hub.hcompany.ai/holotab/routines-and-schedules.md): Record a HoloTab routine by doing the task once, run it from the / menu with new variables, and schedule it to run daily, weekly, or monthly.
- [Autonomy and safety](https://hub.hcompany.ai/holotab/autonomy-and-safety.md): Choose how much HoloTab does without asking, understand what it sees and sends, which sites it refuses, and how to spot and stop a prompt injection.
- [HoloDesktop CLI](https://hub.hcompany.ai/holo-desktop-cli/introduction.md): HoloDesktop CLI runs H's Holo computer-use model on your own Mac, Windows, or Linux desktop from the terminal, MCP, ACP, A2A, or Python.
- [HoloDesktop CLI quickstart](https://hub.hcompany.ai/holo-desktop-cli/getting-started/quickstart.md): Install HoloDesktop CLI on macOS, Windows, or Linux, sign in or point it at a local model, run a first desktop task, and stop it.
- [Hosted or local models](https://hub.hcompany.ai/holo-desktop-cli/getting-started/hosted-or-local-models.md): Choose how HoloDesktop CLI calls a model: hosted mode on H's Models API, or local mode against your own llama.cpp or vLLM server.
- [Architecture](https://hub.hcompany.ai/holo-desktop-cli/architecture.md): How the HoloDesktop CLI Python client, the hai-agent-runtime, the hosted or local model backend, and the CLI, MCP, ACP, A2A, and Python surfaces fit together.
- [Security and privacy](https://hub.hcompany.ai/holo-desktop-cli/security-and-privacy.md): Understand what HoloDesktop CLI sends to H Company, what stays local, and how hosted and local model modes differ.
- [Use HoloDesktop CLI from an agent host](https://hub.hcompany.ai/holo-desktop-cli/integrations/agent-hosts.md): Let Claude Code, Cursor, Codex, and other hosts hand desktop work to HoloDesktop CLI over MCP, ACP, or A2A.
- [Use HoloDesktop CLI as a skill](https://hub.hcompany.ai/holo-desktop-cli/integrations/use-as-skill.md): Understand the HoloDesktop CLI skill that gets installed into supported agent hosts.
- [Customize with skills, memories, and rules](https://hub.hcompany.ai/holo-desktop-cli/how-to/customize.md): Shape how HoloDesktop CLI behaves across every run with optional files in ~/.holo.
- [Debug a failed run](https://hub.hcompany.ai/holo-desktop-cli/how-to/debug-failed-run.md): Use doctor, logs, permissions, and model checks to debug a HoloDesktop CLI run.
- [Embed with Python](https://hub.hcompany.ai/holo-desktop-cli/how-to/embed-with-python.md): Use the HoloDesktop CLI Python agent client to start the runtime, create desktop sessions, stream events, and cancel from your own code.
- [CLI reference](https://hub.hcompany.ai/holo-desktop-cli/reference/cli.md): HoloDesktop CLI commands and options: holo run, stop, mcp, acp, serve, guard, install, login, whoami, doctor, and their flags.
- [Environment variables](https://hub.hcompany.ai/holo-desktop-cli/reference/environment-variables.md): Every environment variable HoloDesktop CLI and hai-agent-runtime read: API key, local model endpoint, port, token, runs dir, download overrides.
- [Paths and files](https://hub.hcompany.ai/holo-desktop-cli/reference/paths-and-files.md): Local files, logs, runtime cache, skills, and run artifacts used by HoloDesktop CLI.
- [Examples](https://hub.hcompany.ai/holo-desktop-cli/examples/index.md): Complete HoloDesktop CLI workflows: UI QA from Claude Code, expense reports from receipts, and more.
- [Find and fix a UI bug with Claude Code](https://hub.hcompany.ai/holo-desktop-cli/examples/claude-code-ui-qa.md): Use HoloDesktop CLI through MCP from Claude Code to test a running app, fix the source, and verify the UI.
- [Automate an expense report from the CLI](https://hub.hcompany.ai/holo-desktop-cli/examples/expense-report-automation.md): Run HoloDesktop CLI directly from the command line to read receipts, update a spreadsheet, draft an email, and verify the result.
- [Docs MCP](https://hub.hcompany.ai/docs-mcp-server.md): Connect Claude, Cursor, VS Code, and other MCP clients to search and read the H Platform documentation.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.