# Antaeus

> Antaeus is an early, Apache-2.0 open-source project for semantic policy
> enforcement: turning human-written business policies into versioned,
> testable, inspectable decisions (allow, review, deny, or failure) that
> applications can consume. The Go core and CLI work locally without
> an account, with a credential-free quickstart and experimental semantic
> evaluators for OpenAI (your own key) and self-hosted CLM. Releases install
> with Homebrew, go install, or a container image. The hosted beta supports API keys, managed policies, and one-time prepaid credit, with shared capacity and no production SLA.

Antaeus is for bounded decisions that need interpretation of language,
intent, context, or exceptions, such as whether a submission complies with a
content or onboarding policy. It does not replace authentication,
authorization, deterministic business rules, transactions, or application
side effects: if a rule can be written safely as ordinary code, it does not
belong in Antaeus. Evaluators only report whether each policy condition
applies; Antaeus applies the policy's own outcomes deterministically.
Decisions produced by the profile runner record the evaluator, model, and
attempt count.

## Install

- Homebrew (macOS, Linux): `brew install antaeusio/tap/antaeus`
- Go 1.26 or newer: `go install github.com/antaeusio/antaeus/cmd/antaeus@latest`
- Container: `docker run --rm ghcr.io/antaeusio/antaeus version`
- [Release archives](https://github.com/antaeusio/antaeus/releases/latest): macOS, Linux, and Windows with SHA256SUMS and build-provenance attestations

## Website

- [Home](https://antaeus.io/): what semantic policy decisions are and the Antaeus lifecycle
- [What is semantic policy enforcement?](https://antaeus.io/semantic-policy-enforcement): definition, how it differs from authorization and rule engines, how decisions are made, when not to use it, and FAQ
- [Open source](https://antaeus.io/open-source): what works today, the local workflow, and maturity
- [Research](https://antaeus.io/research): the Semantic Policy Enforcement research question and evidence standard
- [About](https://antaeus.io/about): project purpose, principles, and stewardship
- [Privacy](https://antaeus.io/privacy): the website's data footprint
- [Terms](https://antaeus.io/terms): terms for the hosted beta and prepaid credit
- [Security](https://antaeus.io/security): vulnerability reporting, safe harbor, and security posture

## Blog

- [Blog](https://antaeus.io/blog): release notes and engineering guides for semantic policy decisions
- [An Amazon Policy in JSON: Know What an Allow Covers](https://antaeus.io/blog/amazon-listing-policy-scope): See the complete two-rule Amazon policy JSON, evaluate a listing, and use its Decision without mistaking selected checks for full compliance.
- [An eBay Policy in JSON: From Listing to Decision](https://antaeus.io/blog/ebay-chance-listings-policy): See the complete eBay policy JSON, run a listing through Antaeus, inspect the Decision, and build the same request for your application.
- [A Facebook Policy in JSON: Preserve the Exception](https://antaeus.io/blog/facebook-policy-exceptions): See the complete Facebook policy JSON and compare Decisions for unpaid cross-promotion, engagement trading and missing context.
- [An X Policy in JSON: Test Reporting Intent](https://antaeus.io/blog/x-reporting-policy-intent): See the complete X policy JSON and compare Decisions for good-faith reporting, an explicit abusive plan and a request with missing context.
- [How to Turn Terms of Use into a Content Policy with Antaeus](https://antaeus.io/blog/terms-of-use-content-policy): Explore Antaeus's public source catalog, turn selected terms into testable rules, and send content to the Antaeus API for a decision.
- [How to Build Custom Content Moderation Rules for a Marketplace](https://antaeus.io/blog/custom-marketplace-content-moderation): Build custom marketplace moderation rules with a versioned policy, explicit decisions, test cases, and clear handling for uncertainty and failures.
- [OpenAI Moderation API and Custom Policies: When to Use Each](https://antaeus.io/blog/openai-moderation-custom-policies): Compare predefined moderation categories with custom rules, then send content and your policy to Antaeus for an explicit decision.
- [Antaeus v0.6.0 adds provider token accounting](https://antaeus.io/blog/antaeus-v0-6-0): Opt-in Drex usage reporting preserves token accounting across retries and fallback while keeping unknown usage distinct from zero.
- [Antaeus v0.5.0 brings the engine to WebAssembly](https://antaeus.io/blog/antaeus-v0-5-0): An experimental WebAssembly build lets JavaScript hosts evaluate policies with the Antaeus engine and their own fetch transport.
- [Antaeus v0.4.0 adds the hosted Drex evaluator](https://antaeus.io/blog/antaeus-v0-4-0): Evaluate semantic policies through the Drex API with System One adapter 0.3.0 and provider-aware retry waits.
- [Antaeus v0.2.0 adds self-hosted CLM evaluation](https://antaeus.io/blog/antaeus-v0-2-0): A second experimental semantic evaluator supports your own CLM server, while container images offer another way to run the engine.
- [Antaeus v0.3.0 adds CPU-only semantic evaluation](https://antaeus.io/blog/antaeus-v0-3-0): The System One adapter gains an Antaeus provider for the open-source NLI server, with existing CLM profiles retained.
- [Antaeus v0.1.0 introduces the local policy engine](https://antaeus.io/blog/antaeus-v0-1-0): The first public release brings versioned policies, explicit decisions, a credential-free quickstart, and an experimental OpenAI evaluator.

## Documentation

- [Hosted API docs](https://antaeus.io/docs): runnable quickstart, authentication, policy sync, limits, errors, idempotency, and prepaid pricing
- [MCP server](https://antaeus.io/docs/mcp): API-key setup and check_intent for proposed actions against an approved policy
- [API reference](https://antaeus.io/docs/api-reference): generated from the released portable OpenAPI contract
- [API limits](https://antaeus.io/docs/limits): workspace allowances, shared capacity, sizes, and concurrency
- [Prepaid pricing](https://antaeus.io/docs/pricing): one-time credit, usage rates, opt-in, and refunds
- [Hosted API plain-text guide](https://antaeus.io/hosted-api.md): the same hosted guide in Markdown
- [Agent integration prompt](https://antaeus.io/integrate-antaeus.txt): generic instructions; the dashboard supplies the configured policy reference

- [Repository](https://github.com/antaeusio/antaeus): source code, issues, and license
- [Quickstart](https://github.com/antaeusio/antaeus/blob/main/docs/quickstart.md): credential-free local validation, evaluation, and regression tests
- [Policy authoring](https://github.com/antaeusio/antaeus/blob/main/docs/policy-authoring.md): policy artifact fields, syntax, and digest identity
- [Reading a Decision](https://github.com/antaeusio/antaeus/blob/main/docs/decisions.md): outcomes, failures, and fallback boundaries
- [Portable contracts](https://github.com/antaeusio/antaeus/blob/main/contracts/README.md): JSON Schemas, OpenAPI description, and conformance fixtures

## Optional

- [OpenAI adapter](https://github.com/antaeusio/antaeus/blob/main/docs/openai-adapter.md): the experimental semantic evaluator, its request boundary, and failures
- [Evaluator adapters](https://github.com/antaeusio/antaeus/blob/main/docs/evaluator-adapters.md): the provider-neutral evaluator boundary
- [Profile execution](https://github.com/antaeusio/antaeus/blob/main/docs/profile-execution.md): retries, deadlines, fallback routing, and traces
- [Contributing](https://github.com/antaeusio/antaeus/blob/main/CONTRIBUTING.md): how to contribute
- [Security policy](https://github.com/antaeusio/antaeus/blob/main/SECURITY.md): private vulnerability reporting