# Averta

## Docs

- [Current limits](https://docs.averta.io/anthropic/current-limits.md): Current Anthropic wrapper limits and fail-closed behavior.
- [Messages API](https://docs.averta.io/anthropic/messages-api.md): Guard Anthropic `messages.create(...)` calls with Averta.
- [Output checks](https://docs.averta.io/anthropic/output-checks.md): Evaluate Anthropic final text before your application returns it.
- [Anthropic](https://docs.averta.io/anthropic/overview.md): Use `wrapAnthropic(...)` to protect Anthropic Messages API calls.
- [Quickstart](https://docs.averta.io/anthropic/quickstart.md): Protect an Anthropic Messages agent loop with Averta decisions.
- [Streaming](https://docs.averta.io/anthropic/streaming.md): Understand Anthropic streaming support in the current wrapper.
- [Tool results](https://docs.averta.io/anthropic/tool-results.md): Evaluate Anthropic `tool_result` content before it goes back to the model.
- [Tool use](https://docs.averta.io/anthropic/tool-use.md): Filter Anthropic tools and check model-requested `tool_use` blocks.
- [Create decision](https://docs.averta.io/api-reference/endpoint/create-decision.md): Call Averta's runtime decision endpoint.
- [Get decision](https://docs.averta.io/api-reference/endpoint/get-decision.md): Decision retrieval is handled through dashboard events today.
- [List decisions](https://docs.averta.io/api-reference/endpoint/list-decisions.md): Decision history is reviewed through dashboard events today.
- [API Reference](https://docs.averta.io/api-reference/introduction.md): The Averta runtime decision API.
- [Choose an integration](https://docs.averta.io/choose-integration.md): Pick the Averta integration path that matches your provider and control level.
- [Checkpoints](https://docs.averta.io/concepts/checkpoints.md): The runtime decision points Averta uses in an agent loop.
- [Decisions](https://docs.averta.io/concepts/decisions.md): Understand Averta decision outcomes and how SDK wrappers handle them.
- [Fail-closed behavior](https://docs.averta.io/concepts/fail-closed-behavior.md): How Averta SDK wrappers handle unsupported or unsafe execution paths.
- [Images and rich media](https://docs.averta.io/concepts/images-and-rich-media.md): How Averta handles images, files, audio, and other non-text inputs.
- [How Averta works](https://docs.averta.io/concepts/overview.md): The runtime model Averta uses to protect agent execution.
- [Policies](https://docs.averta.io/concepts/policies.md): How Averta policies control runtime checkpoint decisions.
- [Request context](https://docs.averta.io/concepts/request-context.md): Correlate Averta decisions across conversations, requests, and traces.
- [Decision client](https://docs.averta.io/custom/decision-client.md): Create a low-level Averta client for custom provider adapters.
- [Output checkpoint](https://docs.averta.io/custom/output-checkpoint.md): Evaluate final model output in a custom provider adapter.
- [Custom integrations](https://docs.averta.io/custom/overview.md): Build a provider adapter or call the raw API when wrappers are not enough.
- [Raw API integration](https://docs.averta.io/custom/raw-api.md): Call Averta's runtime decision API directly from any programming language.
- [Request checkpoint](https://docs.averta.io/custom/request-checkpoint.md): Use the low-level SDK to evaluate provider input before the upstream call.
- [Tool-call checkpoint](https://docs.averta.io/custom/tool-call-checkpoint.md): Evaluate model-requested tool calls before your app executes them.
- [Tool exposure](https://docs.averta.io/custom/tool-exposure.md): Filter provider tools after a low-level Averta decision.
- [Tool results](https://docs.averta.io/custom/tool-results.md): Evaluate returned tool content with the low-level SDK before it goes back to the model.
- [API keys](https://docs.averta.io/dashboard/api-keys.md): Create Averta runtime credentials, attach policies, and use them from SDK wrappers.
- [Events](https://docs.averta.io/dashboard/events.md): Investigate runtime decisions emitted by SDK wrappers and custom integrations.
- [Dashboard overview](https://docs.averta.io/dashboard/overview.md): Use the Averta dashboard to monitor decisions, tune policies, manage keys, and administer your organization.
- [Policies](https://docs.averta.io/dashboard/policies.md): Create reusable policies, configure request enforcement, and manage policy lifecycle.
- [Settings](https://docs.averta.io/dashboard/settings.md): Manage teammates, roles, invitations, and organization-level administration.
- [Tool exposure policy](https://docs.averta.io/dashboard/tool-exposure-policy.md): Control which tools are visible to the model as request risk and intent conditions change.
- [What is Averta](https://docs.averta.io/index.md): Add runtime security checkpoints to OpenAI and Anthropic agent loops.
- [Chat Completions](https://docs.averta.io/openai/chat-completions.md): Guard OpenAI Chat Completions calls, tools, continuations, and streamed text.
- [Configuration](https://docs.averta.io/openai/configuration.md): Configure the OpenAI wrapper, decision callbacks, request context, and errors.
- [Output checks](https://docs.averta.io/openai/output-checks.md): Evaluate OpenAI final output before your application returns it.
- [OpenAI](https://docs.averta.io/openai/overview.md): Use Averta OpenAI wrappers to protect Responses API and Chat Completions calls.
- [Quickstart](https://docs.averta.io/openai/quickstart.md): Protect an OpenAI agent loop with Averta decisions.
- [Responses API](https://docs.averta.io/openai/responses-api.md): Guard OpenAI Responses API calls, tool loops, and streamed output.
- [Streaming](https://docs.averta.io/openai/streaming.md): Use Averta with supported OpenAI streaming calls.
- [Tool results](https://docs.averta.io/openai/tool-results.md): Evaluate OpenAI tool output before it is sent back to the model.
- [Tools](https://docs.averta.io/openai/tools.md): Understand how Averta filters OpenAI tools and checks model-requested tool calls.
- [Quickstart](https://docs.averta.io/quickstart.md): Wrap an OpenAI client and run your first guarded request.
- [Installation](https://docs.averta.io/sdk/installation.md): Install the Averta SDK package for your provider and runtime.

## OpenAPI Specs

- [openapi](https://docs.averta.io/api-reference/openapi.json)

## Optional

- [Blog](https://www.averta.io/blog)