# HUD

- [Introduction](https://docs.hud.ai/v6/start/index.md): Introduction to HUD: define an environment once, run any agent against it for evaluation, and train models on the trajectories your tasks produce.
- [Quickstart](https://docs.hud.ai/v6/start/quickstart.md): Install the HUD SDK, write your first task, run it against a model, and read the reward returned by the grader - the full quickstart in five minutes.
- [Overview](https://docs.hud.ai/v6/start/overview.md): A high-level tour of HUD's core concepts - environments, tasks, capabilities, agents, runtimes, and graders - and how they connect in the central workflow.
- [Creating Environments](https://docs.hud.ai/v6/guides/creating-an-environment.md): Scaffold a HUD environment, implement env.py with tasks and capabilities, test it locally with the CLI, and deploy it to the HUD platform.
- [Evaluating Agents](https://docs.hud.ai/v6/guides/running-an-eval.md): Point an agent at a HUD taskset, pick a runtime for each rollout, run the evaluation, and read the rewards your graders return for every task.
- [Training Agents](https://docs.hud.ai/v6/guides/training-agents.md): Turn an evaluation into a training loop: fork a model, roll out a taskset in groups, and feed the rewards back to nudge the weights.
- [Environments](https://docs.hud.ai/v6/reference/environment.md): Reference for the HUD Environment object: registering tasks, capabilities, and lifecycle hooks, plus serving an environment over the wire protocol.
- [Tasks & Tasksets](https://docs.hud.ai/v6/reference/tasks.md): Reference for HUD Task and Taskset: defining a single task, expanding it into a dataset, grouping tasks into jobs, and syncing tasksets to the platform.
- [Capabilities](https://docs.hud.ai/v6/reference/capabilities.md): Reference for HUD capabilities: the wire protocols an environment exposes, each capability's factory signature, and the clients agents use to attach.
- [Agents](https://docs.hud.ai/v6/reference/agents.md): Agents in HUD: built-in agents like Claude, OpenAI, and Gemini, the HUD gateway for routing models, and how to wire your own agent into a task run.
- [Runtimes](https://docs.hud.ai/v6/reference/runtime.md): Reference for HUD runtimes: local Docker, remote workers, and the platform runtime that decides where each rollout's environment is started and executed.
- [Graders](https://docs.hud.ai/v6/reference/graders.md): Native HUD graders, comparison helpers, and the native grade combiner for scoring agent runs, with patterns for partial credit and multi-step tasks.
- [Designing tasks](https://docs.hud.ai/v6/reference/advice.md): Design HUD tasks that produce learnable, well-calibrated training signal: prompts, graders, reward shaping, and how to debug noisy or sparse rewards.
- [Training](https://docs.hud.ai/v6/reference/training.md): Reference for HUD training: the TrainingClient API, supported training paths, loss functions, run results, and the hud models CLI for managing forks.
- [Types](https://docs.hud.ai/v6/reference/types.md): Reference for HUD serializable types: Run, Trace, Grade, Step, AgentStep, ToolStep, Citation, Answer, and the typed task input and output shapes.
- [CLI](https://docs.hud.ai/v6/reference/cli.md): Reference for the hud command-line interface: init, dev, run, eval, models, and other commands that span the HUD environment and agent lifecycle.
- [Telemetry and analytics](https://docs.hud.ai/v6/reference/telemetry.md): What the HUD SDK sends and how to turn it off.
- [Extending HUD](https://docs.hud.ai/v6/advanced/extending.md): Go beyond the standard workflow with custom harnesses, richer environments, scaled tasksets, subagents as tools, and multi-turn agent chats in HUD.
- [Robots](https://docs.hud.ai/v6/advanced/robots.md): How HUD models physical robots: capability contracts, bridges that connect to sim or hardware, and the agent harness that drives VLA policies.
- [Experimental](https://docs.hud.ai/v6/experimental/index.md): Supported SDK surfaces whose location or organization may change between feature releases.
- [Harbor interoperability](https://docs.hud.ai/v6/experimental/harbor.md): Adapt Harbor tasks into runnable HUD tasksets and Compose artifacts, or export HUD tasks to Harbor folders.
- [Compose environments](https://docs.hud.ai/v6/experimental/compose.md): Declare a multi-service environment as a standard Compose project and run it unchanged locally, hosted, or on owned infrastructure.
- [Verifier environments](https://docs.hud.ai/v6/experimental/verifier-environments.md): Give authoritative grading its own environment, place the acting and grading substrates independently, and control what crosses the phase boundary.
- [Cookbooks: Agents, RL, and Benchmarks](https://docs.hud.ai/v6/cookbooks/index.md): Complete, runnable cookbook examples you can copy and adapt: coding agents, A2A chat, ops diagnostics, robot benchmarks, RL training, and more.
- [Fireworks Serverless RL](https://docs.hud.ai/v6/cookbooks/fireworks-rl-training.md): Train a Fireworks LoRA adapter on rollouts from a HUD environment, with grouped rewards, checkpoints, evaluation, and a path to hosted agent tasks.
- [RL on Daytona sandboxes](https://docs.hud.ai/v6/cookbooks/daytona-rl.md): Run 256 parallel graded rollouts on Daytona and train on the results.
- [Coding agent](https://docs.hud.ai/v6/cookbooks/coding-agent.md): Run a coding agent inside a HUD shell-and-files environment, with tests as the grader, source code in the SDK repo, and a step-by-step walkthrough.
- [Ops diagnostics](https://docs.hud.ai/v6/cookbooks/ops-diagnostics.md): Walk through an ops diagnostics task where the agent gathers logs, metrics, and traces, integrates evidence, and produces a graded root-cause diagnosis.
- [A2A chat](https://docs.hud.ai/v6/cookbooks/a2a-chat.md): Build a chat task on the HUD A2A orchestrator, serve it over the agent-to-agent protocol, and connect with any A2A-compatible client to test turns.
- [Robot benchmark](https://docs.hud.ai/v6/cookbooks/robot-benchmark.md): Run a vision-language-action policy against a containerized robot simulator on HUD, with episodes graded by task success and reproducible task seeds.
- [Internals](https://docs.hud.ai/v6/internals/index.md): How the SDK code actually runs - the module map and the deep dives, for the people changing it.
- [Walkthrough](https://docs.hud.ai/v6/internals/walkthrough.md): A reading tour of one HUD run, following the pointer through the real code from Taskset.run to a graded reward.
- [Placement](https://docs.hud.ai/v6/internals/placement.md): How a task row finds a server - the Provider contract, Runtime addresses, and the routing patterns they compose into.
- [Control channel](https://docs.hud.ai/v6/internals/control-channel.md): The HUD wire contract: JSON-RPC task sessions and raw capability streams multiplexed over one TCP port.
- [FAQ](https://docs.hud.ai/v6/more/faq.md): Answers to the most common questions about getting started with HUD: installation, environments, tasks, runtimes, graders, agents, and training loops.
- [Migrate to v6](https://docs.hud.ai/v6/more/migrate-v6.md): Migrate v5 HUD environments (scenarios, tools, and MCP serving) to the leaner v6 spec built around tasks and capabilities, with side-by-side examples.
- [Contributing](https://docs.hud.ai/v6/more/contributing.md): Contribute to the HUD Python SDK: fork the repo, set up a dev environment with uv, run tests and ruff, and open a pull request against the main branch.
- [Introduction](https://docs.hud.ai/platform/introduction.md): Run models against the environments you build and replay every graded attempt.
- [Environments](https://docs.hud.ai/platform/environments.md): Deploy an environment, follow its build, configure runtime variables, and run its task templates on HUD.
- [Tasksets](https://docs.hud.ai/platform/tasksets.md): Create a reusable evaluation suite on HUD and run it across models.
- [Evaluations](https://docs.hud.ai/platform/evaluations.md): Launch an evaluation job on HUD, then read its traces, rewards, errors, and usage.
- [Models](https://docs.hud.ai/platform/models.md): Choose gateway models for evaluations and manage team-owned trainable model forks on HUD.
- [Training](https://docs.hud.ai/platform/training.md): Run training from the HUD SDK, then inspect model checkpoints, logs, and evaluation results on the platform.
- [QA agents](https://docs.hud.ai/platform/qa-agents.md): Create reusable model-based reviewers for traces and supported platform resources.
- [Data](https://docs.hud.ai/platform/data.md): Upload files for task inputs and control who on the team sees them.
- [File tracking](https://docs.hud.ai/platform/file-tracking.md): How HUD records workspace changes and rebuilds any file at any point in a trace.
- [Team settings](https://docs.hud.ai/platform/team.md): Manage team access, API keys, rollout secrets, provider credentials, usage, billing, and spend limits.
- [REST API](https://docs.hud.ai/platform/rest-api.md): Call the HUD platform over HTTP - authentication, the objects the API is built around, and every endpoint grouped by what it is for.
- [MCP](https://docs.hud.ai/platform/mcp.md): Connect a coding agent to the HUD platform over MCP - setup, authentication, and every read-only tool for environments, tasksets, jobs, and traces.