Sapl Policy Engine

What SAPL does

SAPL is an authorization engine with a human-readable policy language, streaming decisions that update in real time when context changes, and a dedicated testing DSL with coverage reporting. It runs embedded in your application or as a standalone server.

policy "freeze during peak hours"
deny
    subject.role == "engineer";
    action == "deploy";
    resource.environment == "production";
    <time.localTimeIsBetween("09:00", "17:00", "Europe/Berlin")>;
obligation {
    "type": "notify",
    "channel": "ops-alerts",
    "message": "Deployment blocked: peak hours"
}

Policies can do more than allow or deny. Obligations and advice attach machine-readable instructions to decisions: redact fields, rewrite queries, log access, require human approval, or trigger notifications. The framework enforces them automatically.

Why SAPL

• Streaming authorization. Subscribe to decisions. The PDP pushes updates when attributes change. No polling, no stale decisions.
• Human-readable policies. Not Datalog, not YAML, not XML. Policies read like structured English.
• Testing DSL. The only authorization engine with a dedicated test language. Mock attribute sources, emit streaming changes, assert decision sequences, enforce coverage thresholds in CI.
• Obligations and advice. Every decision can carry structured instructions that the framework executes automatically.
• AI agent authorization. Control tool calls, RAG retrieval, and MCP operations with per-tool policies and human-in-the-loop approval workflows.

Quick start

Try in the browser. The Playground runs entirely in your browser. Write policies, create subscriptions, observe decisions. No install needed.

Try on the command line. Download the sapl binary from the releases page. Write a policy:

policy "Dr. House is allowed to use the MRT!"
permit
    subject == "housemd" & action == "use" & resource == "MRT";

Evaluate it:

sapl decide-once -s '"housemd"' -a '"use"' -r '"MRT"'
# {"decision":"PERMIT"}

sapl decide-once -s '"cuddy"' -a '"use"' -r '"MRT"'
# {"decision":"DENY"}

The Getting Started guide covers the full CLI workflow including streaming decisions and testing.

Add the starter:

<dependency>
    <groupId>io.sapl</groupId>
    <artifactId>sapl-spring-boot-starter</artifactId>
    <version>4.0.0</version>
</dependency>

Annotate a method:

@PreEnforce(action = "'read'", resource = "'patient'")
public Patient getPatient(String id) {
    return repository.findById(id);
}

Write a policy in src/main/resources/policies/:

policy "doctors read patients"
permit
    subject.role == "doctor";
    action == "read";
    resource == "patient";

The Spring scenario walks through a complete application step by step.

Use it with any stack

Test policies like you test code

SAPL is the only authorization engine with a dedicated testing language. Mock attribute sources, emit streaming changes, assert decision sequences, and enforce coverage thresholds in CI.

policy "permit on emergency"
permit
    action == "read" & resource == "time";
    "status".<mqtt.messages> == "emergency";

scenario "decision changes when emergency status changes"
    given
        - attribute "statusMock" "status".<mqtt.messages> emits "emergency"
    when "user" attempts "read" on "time"
    expect permit
    then
        - attribute "statusMock" emits "ok"
    expect not-applicable
    then
        - attribute "statusMock" emits "emergency"
    expect permit;

> sapl test --policy-hit-ratio 100 --condition-hit-ratio 100

  permit_on_emergency.sapltest
    PASS  permit when MQTT status is emergency
    PASS  not-applicable when MQTT status is ok
    PASS  alternating permit and not-applicable with multiple status changes

Tests:    10 passed, 10 total
Coverage: Policy Hit 100.00%  Condition Hit 100.00%

Scenarios

Working demos with code walkthroughs:

• Spring Security -- method-level ABAC in a Spring Boot application
• AI Tool Authorization -- per-tool gating for Spring AI
• RAG Pipeline -- dynamic query rewriting for retrieval-augmented generation
• Human-in-the-Loop -- policy-driven approval workflows for AI tool calls
• MCP Server -- authorize tool calls, resources, and prompts in MCP servers

Get involved

• Playground -- try SAPL policies in the browser, no install needed
• Get Started -- add SAPL to your first project
• Documentation -- language reference, functions, integrations
• Demos -- runnable example projects
• Discord -- questions, ideas, help from the team

IDE and editor support

The SAPL language server provides syntax highlighting, diagnostics, content assist, formatting, and more for .sapl and .sapltest files in any LSP-compatible editor. See the IDE setup guide for IntelliJ, VS Code, Neovim, and other editors.

Compatibility

Contributing

See CONTRIBUTING.md.

Security

See SECURITY.md.

Code of Conduct

See CODE_OF_CONDUCT.md.

License

Apache 2.0. See LICENSE.

Snapshots provide the newest development state. Add the snapshot repository to your build:

Maven

<repositories>
    <repository>
        <id>central-portal-snapshots</id>
        <url>https://central.sonatype.com/repository/maven-snapshots/</url>
        <releases><enabled>false</enabled></releases>
        <snapshots><enabled>true</enabled></snapshots>
    </repository>
</repositories>

Gradle

repositories {
    maven { url = uri("https://central.sonatype.com/repository/maven-snapshots") }
}

Need a Software Bill of Materials? See dependency graph.