# MUTAGENT

- [What is Helix](https://docs.mutagent.io/helix/what-is-helix.md): A coding agent that builds and improves AI agents: spec, build, evaluate, diagnose, optimize.
- [Quickstart](https://docs.mutagent.io/quickstart/overview.md): Install Helix, sign in to Mutagent, and run your first loop in about five minutes.
- [Install Helix](https://docs.mutagent.io/helix/install.md): One command installs Helix, the `helix` binary. No Node, no npm, no checkout.
- [Your first loop](https://docs.mutagent.io/helix/install/first-loop.md): One agent — a contract reviewer — taken from a sentence to a validated spec, a build, and a first verdict.
- [Three ways to run Helix](https://docs.mutagent.io/helix/modes.md): One program, three ways to start it: the full Helix session, a plain coding agent, or the experimental code-first mode. What each gives you and what to type.
- [Agent mode](https://docs.mutagent.io/helix/agent-mode.md): helix agent — Helix as a lean, general-purpose coding agent, without the lifecycle. Interactive, loading one agent definition, or headless with -p.
- [Sessions](https://docs.mutagent.io/helix/sessions.md): What you see when Helix boots — the dashboard, the slash commands, and the working line.
- [How the loop works](https://docs.mutagent.io/helix/how-the-loop-works.md): The five ADLC stages, how they hand off, and the two rules that keep the loop honest.
- [Spec](https://docs.mutagent.io/helix/lifecycle/spec.md): Stage 1: define what your agent is. A guided interview that emits a portable agentspec.yaml.
- [Build](https://docs.mutagent.io/helix/lifecycle/build.md): Stage 2: implement a validated spec in your framework or coding agent, test-first.
- [Evaluate](https://docs.mutagent.io/helix/lifecycle/evaluate.md): Stage 3: score your agent against real traces. Derives criteria, judges each one, and validates the judge before trusting it.
- [Diagnose](https://docs.mutagent.io/helix/lifecycle/diagnose.md): Stage 4: root-cause the failures on evidence and rank the remedies. Failures come from Evaluate; Diagnose applies only the remedies you approve.
- [Optimize](https://docs.mutagent.io/helix/lifecycle/optimize.md): Stage 5: apply an approved remedy and re-evaluate until your criteria pass. You confirm the goal at the start and approve once before the final change.
- [Ship](https://docs.mutagent.io/helix/lifecycle/ship.md): After the loop: ship an agent that passed its checks, then watch it after it deploys. Only you start it.
- [Overview](https://docs.mutagent.io/helix/features.md): The whole capability surface of the `helix` binary, at a glance — sub-agents, agent mode, goals, monitoring, tracing, Prime, and the rest.
- [Sub-agents](https://docs.mutagent.io/helix/features/sub-agents.md): Start sub-agents in parallel, watch each one live in the fleet list, and steer any of them by typing at it.
- [Workflows](https://docs.mutagent.io/helix/features/workflows.md): Fan work out across sub-agents from a small script — saved in your repo, run by name, and watched live as it goes.
- [Goals](https://docs.mutagent.io/helix/features/goals.md): /goal keeps the agent working toward one objective until it is done, blocked, or out of the budget you set.
- [Monitor](https://docs.mutagent.io/helix/features/monitor.md): The agent watches a long-running command in the background and reacts as output arrives, line by line.
- [Trace viewer](https://docs.mutagent.io/helix/features/trace-viewer.md): /trace — a read-only viewer over your session logs: list sessions, open one, walk its trajectory. Available in every mode.
- [Project context](https://docs.mutagent.io/helix/features/project-context.md): AGENTS.md / CLAUDE.md loads into the prompt in every mode, nearest file last, and reaches the agents Helix dispatches.
- [MCP servers](https://docs.mutagent.io/helix/features/mcp.md): Connect your own MCP servers to Helix. Their tools reach the model directly.
- [Helix Cloud](https://docs.mutagent.io/helix/cloud/overview.md): Run Helix in a cloud sandbox. Store tool secrets in an Environment, connect your LLM providers, then run a task or keep a session running.
- [Environments](https://docs.mutagent.io/helix/cloud/environments.md): Store the variables and secrets your cloud sessions' tools need in a workspace Environment, and load it into a run with --env.
- [Sandbox providers](https://docs.mutagent.io/helix/cloud/sandbox-providers.md): A sandbox provider is where a cloud sandbox runs. Mutagent runs your sessions in its own sandboxes unless you choose another configured sandbox provider.
- [LLM providers and models](https://docs.mutagent.io/helix/cloud/setup.md): Connect your LLM providers to the workspace so cloud sessions can call your models, and choose the default model a run uses.
- [Run a task](https://docs.mutagent.io/helix/cloud/agent-runs.md): Give Helix a task in a cloud sandbox and print the result: the full Helix session, Prime, or your own agent definition, with text or JSON output.
- [Sessions](https://docs.mutagent.io/helix/cloud/sessions.md): Keep an interactive cloud Helix session running with --mode rpc, then list, watch, steer, stop, checkpoint, and restore it with mutagent helix session.
- [Run managed agents on Helix Cloud](https://docs.mutagent.io/helix/cloud/managed-agents.md): A managed agent is an agent you write as an agent.md folder with its own tools, skills and files, deploy to your workspace, and run on Helix Cloud by its address (`@slug`).
- [Troubleshooting cloud runs](https://docs.mutagent.io/helix/cloud/troubleshooting.md): What the errors from mutagent helix mean and what to do: models, launch arguments, sessions, idle sandboxes, Environments, and LLM provider mirroring.
- [Gateway overview](https://docs.mutagent.io/helix/gateway/overview.md): The gateway starts Helix runs from events and schedules: a GitHub mention, an opened issue, a Slack message, a webhook, or a time of day.
- [Connecting GitHub and Slack](https://docs.mutagent.io/helix/gateway/connect.md): Connect each provider once for your organization, choose the repositories the gateway works on, and know why signing in with GitHub is not enough.
- [Triggers and routines](https://docs.mutagent.io/helix/gateway/triggers-and-routines.md): The event kinds a trigger can listen for, what each one fires on, and how routines run an instruction on a schedule.
- [Commands](https://docs.mutagent.io/helix/reference/commands.md): The Helix command roster. Every stage has a command, but you can drive it all in plain English.
- [Config](https://docs.mutagent.io/helix/reference/config.md): The Helix configuration reference: file location, the blocks, and the keys you set most.
- [Traces](https://docs.mutagent.io/helix/traces.md): Helix judges your agent against the runs it already logs. Here's what counts as a trace and where it reads from.
- [FAQ](https://docs.mutagent.io/helix/reference/faq.md): Common questions about Helix: scoring, data, modes, provider keys, and removal.
- [CLI overview](https://docs.mutagent.io/cli/overview.md): The Mutagent CLI signs you in, installs Helix, runs Helix and your agents in the cloud, and manages your workspace's LLM providers and Environments.
- [CLI installation](https://docs.mutagent.io/cli/installation.md): Install the Mutagent CLI, sign in (in a terminal, from a coding agent, or with an API key), choose a workspace, and check that it worked.
- [Commands](https://docs.mutagent.io/cli/commands.md): The Mutagent CLI commands, grouped by what they are for, and the flags that work with every command.
- [Errors and exit codes](https://docs.mutagent.io/cli/errors.md): What the Mutagent CLI prints when a command fails, the JSON error object, the exit codes, and why delete commands need --force.
- [Scripts, CI and coding agents](https://docs.mutagent.io/cli/json.md): Run the Mutagent CLI without a person at the keyboard: sign in with an API key, read --json output, branch on exit codes, and let a coding agent drive it.
- [mutagent login](https://docs.mutagent.io/cli/commands/login.md): Sign in to Mutagent, check your sign-in, and choose the workspace and organization commands act on.
- [mutagent usage, feedback](https://docs.mutagent.io/cli/commands/account.md): Count your workspace's LLM providers and your workspaces, and send product feedback to the Mutagent team.
- [mutagent providers](https://docs.mutagent.io/cli/commands/providers.md): Add, copy, test, and manage the LLM providers whose API keys your cloud sessions use.
- [mutagent env](https://docs.mutagent.io/cli/commands/env.md): Create and manage workspace Environments: the variables and secrets your cloud sessions' tools need.
- [mutagent helix](https://docs.mutagent.io/cli/commands/helix.md): Run Helix in a cloud sandbox: the full Helix session, Prime, or your own agent, on a task or as a live session.
- [mutagent helix session](https://docs.mutagent.io/cli/commands/helix-session.md): List, watch, send commands to, stop, checkpoint, and restore Helix Cloud sessions from any terminal.
- [mutagent helix models](https://docs.mutagent.io/cli/commands/helix-models.md): List the models your Helix Cloud sessions can use, and set the default model a run uses when it names none.
- [mutagent helix version, doctor, smoke](https://docs.mutagent.io/cli/commands/helix-diagnostics.md): Check which Helix version cloud sandboxes run, and run Helix's own diagnostics and smoke test in a cloud sandbox.
- [mutagent sandbox](https://docs.mutagent.io/cli/commands/sandbox.md): List the sandbox providers and presets a Helix Cloud run can name with --sandbox-provider and --preset.
- [mutagent agent](https://docs.mutagent.io/cli/commands/agent.md): Check, pack, deploy, activate, list, inspect, retire, and locally run agents written as an agent.md folder.
- [mutagent reports, findings, triage](https://docs.mutagent.io/cli/commands/reports.md): Store Helix's evaluation and diagnosis reports in your workspace, track the findings they raise, and file problems about an agent.
- [mutagent gateway](https://docs.mutagent.io/cli/commands/gateway.md): Connect GitHub and Slack, link repositories, and configure the triggers and routines that start Helix runs.
- [mutagent install](https://docs.mutagent.io/cli/commands/install.md): Install the Helix binary on your machine.
- [mutagent init, skills, hooks](https://docs.mutagent.io/cli/commands/project-setup.md): Connect a project to Mutagent, install the CLI skill for coding agents, and send Claude Code session telemetry as traces.
- [What is Mutagent](https://docs.mutagent.io/introduction.md): Mutagent takes an AI agent through one loop: spec, build, evaluate, diagnose, optimize. Here is where each part of the product fits.
- [Accounts, organizations and workspaces](https://docs.mutagent.io/get-started/accounts.md): How your Mutagent account is organized, which workspace a command runs in, and how to switch.
- [Using the web app](https://docs.mutagent.io/get-started/web-app.md): Start a session, follow it as it works, and manage your workspace at app.mutagent.io.
- [API keys](https://docs.mutagent.io/quickstart/api-keys.md): Organization keys and workspace keys: which to create, where to create them, and how to use one in CI.
- [What's new](https://docs.mutagent.io/get-started/whats-new.md): Dated changes to Helix, the mutagent CLI and the platform, newest first.
- [Cloud sandboxes](https://docs.mutagent.io/platform/sandbox/overview.md): What a cloud sandbox is on the platform, what the platform manages for it, and what you control.
- [Sandbox providers](https://docs.mutagent.io/platform/sandbox/providers.md): A sandbox provider runs cloud sandboxes. What it decides, what it does not, and how a run picks one.
- [Agents on the platform](https://docs.mutagent.io/platform/agents/overview.md): Two ways to run agents on the platform: Helix in a cloud sandbox, and managed agents deployed from an agent.md folder. Where each is documented.
- [How managed agents work](https://docs.mutagent.io/platform/managed-agents/overview.md): Write an agent as files, check and run it on your machine, deploy it to your workspace, and run it on Helix Cloud by its address (`@<slug>`).
- [Quickstart: build and deploy an agent](https://docs.mutagent.io/platform/managed-agents/quickstart.md): Go from an empty folder to a deployed agent that runs on Helix Cloud: one tool, one skill, one file, a local run, a deploy, a new revision and a rollback.
- [agent.md reference](https://docs.mutagent.io/platform/managed-agents/agent-md.md): Every field of a managed agent's agent.md, the folder around it, what check and pack validate and produce, and how the platform uses each field.
- [Tools, skills and files](https://docs.mutagent.io/platform/managed-agents/tools-and-skills.md): How to write a tool module, add Helix built-in tools, write a skill, package files, declare secrets, and check, run, pack and deploy the agent.
- [Deployment](https://docs.mutagent.io/platform/managed-agents/deployment.md): How a managed agent moves through deploy, activate, run, retire and archive, and what each step checks and refuses.
- [LLM providers](https://docs.mutagent.io/platform/providers/overview.md): Store your own model keys in a workspace for cloud runs. Each LLM provider configuration belongs to one workspace.
- [LLM provider setup](https://docs.mutagent.io/platform/providers/setup.md): Add an LLM provider to your workspace, field by field, for every catalog entry.
- [SDK](https://docs.mutagent.io/sdk/overview.md): Use your Mutagent account from TypeScript or Python: workspaces, LLM providers, Environments, Helix Cloud sessions and managed agents.
- [Installation](https://docs.mutagent.io/sdk/typescript/installation.md): Install the Mutagent TypeScript SDK and make your first call.
- [Configuration](https://docs.mutagent.io/sdk/typescript/configuration.md): Configure the Mutagent TypeScript SDK client.
- [LLM providers](https://docs.mutagent.io/sdk/typescript/providers.md): List, add and test the workspace's LLM providers with the TypeScript SDK.
- [Environments and sandbox providers](https://docs.mutagent.io/sdk/typescript/environments.md): List, create, update and delete workspace Environments, and list sandbox providers and presets, with the TypeScript SDK.
- [Helix Cloud and managed agents](https://docs.mutagent.io/sdk/typescript/helix-cloud.md): Launch Helix on Helix Cloud, run managed agents, control running sessions, and deploy managed agents with the TypeScript SDK.
- [Installation](https://docs.mutagent.io/sdk/python/installation.md): Install the Mutagent Python SDK and make your first call.
- [Configuration](https://docs.mutagent.io/sdk/python/configuration.md): Configure the Mutagent Python SDK client.
- [LLM providers](https://docs.mutagent.io/sdk/python/providers.md): List, add and test the workspace's LLM providers with the Python SDK.
- [Environments and sandbox providers](https://docs.mutagent.io/sdk/python/environments.md): List, create, update and delete workspace Environments, and list sandbox providers and presets, with the Python SDK.
- [Helix Cloud and managed agents](https://docs.mutagent.io/sdk/python/helix-cloud.md): Launch Helix on Helix Cloud, run managed agents, control running sessions, and deploy managed agents with the Python SDK.
- [Integrations](https://docs.mutagent.io/integrations/overview.md): How Helix reads your traces, writes approved fixes, and connects to GitHub, Slack and the cloud platforms around your agents.
- [Supported integrations](https://docs.mutagent.io/get-started/supported-integrations.md): Every LLM provider, sandbox provider, trace source, apply target and channel Mutagent works with, and where each one is set up.
- [Local trace sources](https://docs.mutagent.io/integrations/sources/local.md): Read your agent's runs from local transcripts and JSONL exports — no cloud account needed.
- [OpenTelemetry sources](https://docs.mutagent.io/integrations/sources/otel.md): Read traces from Langfuse or from OpenTelemetry (OTLP/JSON) span files, or connect Langfuse and OpenObserve to your workspace.
- [Dash0](https://docs.mutagent.io/integrations/sources/dash0.md): Read your agent's traces from Dash0. Coming soon.
- [Markdown agent targets](https://docs.mutagent.io/integrations/targets/harness.md): Apply approved fixes to markdown agents — Claude Code, Codex, Cursor, OpenCode and skill directories — via GitHub.
- [Code targets](https://docs.mutagent.io/integrations/targets/code.md): Apply approved fixes to code-based agents — Mastra and the Claude Agent SDK — via GitHub. DeepAgents and Pydantic AI are coming soon.
- [GitHub](https://docs.mutagent.io/integrations/github.md): Ship approved fixes as pull requests from your checkout, and start Helix runs from GitHub issues, mentions and failed CI.
- [Slack](https://docs.mutagent.io/integrations/slack.md): Mention the Mutagent Slack app to start a Helix run, and get the reply in the thread.
- [Cloud deployment](https://docs.mutagent.io/integrations/cloud-deployment.md): Where your agent runs in the cloud: Mutagent managed agents, or a platform you already use.


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