# Altrum

> The product guide for Altrum: govern your AI systems from assessment to every request they make.

- [Introduction](https://docs.altrum.ai/index.md): What Altrum does for your organization's AI, and where to start reading.
- [How Altrum works](https://docs.altrum.ai/how-altrum-works.md): Understand the six stages of AI governance in Altrum and the pieces that make each one work.
- [Quickstart](https://docs.altrum.ai/quickstart.md): Go from your invitation to an AI system with live guardrails, approved and ready for traffic.
- [Start by role](https://docs.altrum.ai/start-by-role.md): Find the pages that matter for your job: governance and risk, security, building with AI, running Altrum, or auditing.
- [Home](https://docs.altrum.ai/home.md): Read your organization's AI governance at a glance: what needs you, how compliant your AI systems are, and what your AI traffic is doing.
- [Sign in](https://docs.altrum.ai/sign-in.md): Accept an invitation, sign in with a password or Google, reset a forgotten password, and know when you are signed out.
- [Account and organization](https://docs.altrum.ai/account-and-organization.md): Change your own details, theme and email settings, and see or edit your organization's profile.
- [Members and roles](https://docs.altrum.ai/members-and-roles.md): Understand the five roles, see who can do what, and invite, change, disable and remove the people in your organization.
- [Plans and modules](https://docs.altrum.ai/plans-and-modules.md): Understand what your plan includes, which screens each module adds, and what happens when a plan ends.
- [Discover](https://docs.altrum.ai/discover.md): Find the AI your organization uses that nobody registered, see how much of your traffic is governed, and register it, attach it or dismiss it.
- [AI systems](https://docs.altrum.ai/ai-systems.md): Register the AI systems your organization builds or buys, keep their details current, and find everything about each one on its page.
- [Lifecycle](https://docs.altrum.ai/ai-systems/lifecycle.md): Take an AI system from Draft to Live with an approval, pause it, retire it, and see what the go-live checks need at each risk level.
- [Risk rating](https://docs.altrum.ai/ai-systems/risk-rating.md): How Altrum works out an AI system's risk rating and EU AI Act category, how an assessment updates it, and how guardrails lower the residual rating.
- [Projects](https://docs.altrum.ai/projects.md): Group AI systems by team or product, with a lead and members, and see their spend and posture together.
- [Model catalog](https://docs.altrum.ai/model-catalog.md): See every model your provider credentials can serve with its price and use, decide which are approved, and make the AI Gateway serve approved models only.
- [MCP servers](https://docs.altrum.ai/mcp-servers.md): Give each MCP server an owner and a status, see what each of its tools can do, and block a server so no key can use its tools.
- [Assessments](https://docs.altrum.ai/assessments/overview.md): Understand what an assessment of an AI system does, who can run one, and how it leads to live guardrails.
- [Run an assessment](https://docs.altrum.ai/assessments/run.md): Answer the assistant's questions about an AI system, attach documents, pause and come back, and submit your answers.
- [Assessment results](https://docs.altrum.ai/assessments/results.md): Review the risks, controls and guardrails an assessment found, add risks it did not find, and move on to configure guardrails.
- [Guardrails](https://docs.altrum.ai/guardrails/overview.md): Understand what a guardrail checks, what it does when it fires, and how a draft becomes the live version.
- [Guardrail types](https://docs.altrum.ai/guardrails/types.md): Reference for the eight guardrails: what each looks for, what you set, and its limits.
- [Configure guardrails](https://docs.altrum.ai/guardrails/configure.md): Switch guardrails on, set what they check, save a draft, and apply it as the live version or complete an assessment.
- [Guardrail suggestions](https://docs.altrum.ai/guardrails/suggestions.md): Use the settings the assistant suggests from an assessment as a starting point for an AI system's guardrails.
- [Test guardrails](https://docs.altrum.ai/guardrails/test.md): Check sample messages against an AI system's saved guardrail draft before you apply it.
- [Red-team runs](https://docs.altrum.ai/guardrails/red-team.md): Send a built-in set of attacks at an AI system's live guardrails, or your organization's, and see what got through and the setting that would catch it.
- [Versions](https://docs.altrum.ai/guardrails/versions.md): See every guardrail version applied to an AI system, and roll back to an earlier one.
- [Organization guardrails](https://docs.altrum.ai/guardrails/organization.md): Guardrails for AI Gateway traffic that does not belong to a registered AI system.
- [Reporting](https://docs.altrum.ai/reports.md): See what your AI traffic costs and where the month is heading, how often your guardrails fire, and which MCP tools are used.
- [Alerts](https://docs.altrum.ai/alerts.md): Get told when a budget, a guardrail, your traffic or an AI system needs you, in the bell and by email, and choose the thresholds.
- [Incidents](https://docs.altrum.ai/incidents.md): Track a problem with an owner, a severity, a status and a timeline, open it from an alert, a report or a request, and record a serious incident under the EU AI Act.
- [Posture](https://docs.altrum.ai/ai-systems/posture.md): See what an AI system's risk rating asks of it, which requirements it meets, and close each gap from its Overview.
- [Compliance reports](https://docs.altrum.ai/compliance.md): Gather the evidence an auditor asks for, for one AI system or the whole organization over a period, then print it or export it as CSV.
- [Access reviews](https://docs.altrum.ai/access-reviews.md): Review every key your organization holds, with when each was last used and what looks wrong, keep or revoke each one, and keep the decisions as evidence.
- [Audit trail](https://docs.altrum.ai/audit-trail.md): See who changed what in your organization, filter the changes, and export them.
- [AI Gateway](https://docs.altrum.ai/ai-gateway/overview.md): Send your applications' and coding agents' model traffic through Altrum, so every request is governed.
- [Send your first request](https://docs.altrum.ai/ai-gateway/quickstart.md): Store a provider credential, create a virtual key, and send your first governed request through the AI Gateway.
- [Provider credentials](https://docs.altrum.ai/ai-gateway/provider-credentials.md): Store your organization's own model provider credentials in Altrum, or let callers send their own.
- [Virtual keys](https://docs.altrum.ai/ai-gateway/virtual-keys.md): Create the keys your applications and people use to send requests through the AI Gateway, and attach them to AI systems.
- [Models and routing](https://docs.altrum.ai/ai-gateway/models-and-routing.md): Choose which models each AI system and key may use, share traffic across credentials, and fall back when a provider fails.
- [SDKs](https://docs.altrum.ai/ai-gateway/sdks.md): Point the OpenAI, Anthropic and Google SDKs at the AI Gateway by changing the base URL and the key.
- [Coding agents](https://docs.altrum.ai/ai-gateway/coding-agents.md): Run Claude Code, Codex CLI, Gemini CLI and Cursor through the AI Gateway with your personal key.
- [MCP tools](https://docs.altrum.ai/ai-gateway/mcp.md): Register your organization's MCP servers in Altrum, control which tools each AI system and key can use, and use them from coding agents and applications.
- [Prompts](https://docs.altrum.ai/ai-gateway/prompts.md): Keep your organization's prompts in Altrum, with versions, and use them from any request.
- [Guardrails at runtime](https://docs.altrum.ai/guardrails/at-runtime.md): How the AI Gateway applies live guardrails to every request and answer, and what your application sees when one fires.
- [Budgets and rate limits](https://docs.altrum.ai/ai-gateway/budgets-and-limits.md): Cap spend and traffic for your organization, each AI system and each key, and see every budget in one place.
- [Semantic cache](https://docs.altrum.ai/ai-gateway/semantic-cache.md): Answer repeated and similar requests from a cache, to cut cost and wait time, without crossing governance boundaries.
- [Request logs](https://docs.altrum.ai/ai-gateway/request-logs.md): See every request and tool call through the AI Gateway, and decide whether Altrum keeps their content.
- [OpenTelemetry](https://docs.altrum.ai/ai-gateway/opentelemetry.md): Send traces of your AI Gateway traffic to your own observability tools.
- [Handle errors](https://docs.altrum.ai/ai-gateway/handle-errors.md): Catch the AI Gateway's refusals in your application, retry the ones that pass, and show people a useful message for the rest.
- [Troubleshooting](https://docs.altrum.ai/ai-gateway/troubleshooting.md): Fix the most common problems with requests through the AI Gateway, by the status code or symptom you see.
- [Introduction](https://docs.altrum.ai/api-reference/introduction.md): The AI Gateway API: base URLs, the request formats it accepts, how to name a model, and what it does not serve.
- [Authentication](https://docs.altrum.ai/api-reference/authentication.md): Send a virtual key with every request to the AI Gateway, and, if your organization allows it, your own provider key.
- [Request headers](https://docs.altrum.ai/api-reference/request-headers.md): Headers you can send with a request to the AI Gateway to use a saved prompt, steer the cache, keep content out of logs or set your own request reference.
- [Response headers](https://docs.altrum.ai/api-reference/response-headers.md): Headers the AI Gateway adds to every reply: the request reference, the trace, and which provider and model answered.
- [Streaming](https://docs.altrum.ai/api-reference/streaming.md): Stream replies from the AI Gateway as server-sent events, and handle an error that arrives after the stream has started.
- [Errors](https://docs.altrum.ai/api-reference/errors.md): Every error the AI Gateway returns: the body shape for each format, the status codes, the refusal codes and what to do about each.
- [Create a chat completion](https://docs.altrum.ai/api-reference/openai/create-a-chat-completion.md): Sends a conversation to a model in the OpenAI chat format. Any supported provider can answer: the reply comes back in the OpenAI format.
- [Create a response](https://docs.altrum.ai/api-reference/openai/create-a-response.md): Sends a request in the OpenAI Responses format. Only create and stream are served: a stored response cannot be fetched, listed or deleted.
- [List models](https://docs.altrum.ai/api-reference/openai/list-models.md): Lists the models your key may use, narrowed to the models your organization approved.
- [Create a message](https://docs.altrum.ai/api-reference/anthropic/create-a-message.md): Sends a request in the Anthropic Messages format. Any supported provider can answer: the reply comes back in the Anthropic format.
- [Count tokens](https://docs.altrum.ai/api-reference/anthropic/count-tokens.md): Counts the tokens a Messages request would use, without sending it to the model.
- [List models](https://docs.altrum.ai/api-reference/anthropic/list-models.md): Lists the models your key may use, narrowed to the models your organization approved.
- [Generate content](https://docs.altrum.ai/api-reference/gemini/generate-content.md): Sends a request in the Gemini format. Any supported provider can answer: the reply comes back in the Gemini format.
- [Stream generated content](https://docs.altrum.ai/api-reference/gemini/stream-generated-content.md): The streaming form of generate content: the answer arrives in parts as it is written.
- [MCP endpoint](https://docs.altrum.ai/api-reference/mcp/mcp-endpoint.md): Altrum's MCP server, for MCP clients and coding agents. It speaks the Model Context Protocol over streamable HTTP (JSON-RPC 2.0) and lists the tools your key may use, across your organization's MCP servers.
- [MCP event stream](https://docs.altrum.ai/api-reference/mcp/mcp-event-stream.md): Opens the server-sent event stream an MCP client listens on.
- [Run a tool](https://docs.altrum.ai/api-reference/mcp/run-a-tool.md): Runs one tool call the model asked for and returns the tool's result, ready to send back to the model. Your guardrails check the arguments before the tool runs and the result before it returns.
- [Support](https://docs.altrum.ai/support.md): Raise a support ticket from inside Altrum and follow it by email.
- [Security and data](https://docs.altrum.ai/security-and-data.md): Where Altrum keeps your data, who can see it, and how access is protected.
- [Error messages](https://docs.altrum.ai/reference/errors.md): Look up a message Altrum showed you, find out what it means and what to do.
- [Emails](https://docs.altrum.ai/reference/emails.md): The emails Altrum sends, who receives each one, and when.
- [Glossary](https://docs.altrum.ai/glossary.md): What the words in Altrum mean.
- [Changelog](https://docs.altrum.ai/changelog.md): What changed in Altrum, release by release, newest first.

## OpenAPI Specs

- [openapi](/api-reference/openapi.json)


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