Govern AI agent tool calls
Let an agent decide what it wants to do while RBEK decides what may actually execute.
Stable CLI 0.1.5
Governed execution for developers.
RBEK is the governed execution layer for AI agents, workflows and software. Install the CLI, define an explicit workflow, authorize execution and produce verifiable evidence of what happened.
curl -fsSL https://releases.rbekplatform.com/cli/stable/install.sh | bash
Public command:
rbek · Compatibility: rbek-cli
What you can do
Put a governed boundary around execution.
RBEK gives developers an explicit control point between software intent and real execution.
Execute controlled workflows
Define inspectable project and workflow files before execution begins.
Call external APIs safely
Route external execution through controlled provider boundaries instead of bypassing governance.
Apply policy before execution
Default-deny admission makes authorization explicit before a governed action is allowed to run.
Produce execution receipts
Capture evidence about the governed execution and verify the resulting receipt.
Integrate providers
Connect real systems through provider adapters while keeping execution authority inside RBEK.
First 5 minutes
Install. Initialize. Run.
The first workflow stays intentionally small so you can inspect the developer surface before connecting external systems.
export PATH="$HOME/.local/bin:$PATH"
mkdir hello-rbek
cd hello-rbek
rbek init
rbek run
Expected first execution:
Hello from RBEK
1. IntentAgent or workflow requests an action
2. PolicyRBEK evaluates authorization
3. GateExecution is allowed or denied
4. ExecuteProvider boundary performs the action
5. ProveReceipt and evidence are produced
Real use case
Govern an AI agent tool call.
A real AI agent requests an external action. RBEK applies policy, executes a real third-party HTTPS call through a governed boundary, and produces verifiable proof.
AI Agent Governance
See the complete public example using a real agent, a real external API and RBEK-controlled execution.
Core concepts
The pieces behind governed execution.
Policy
Rules that determine whether a requested action may proceed.
Execution boundary
The controlled boundary through which authorized work is actually performed.
Provider
An adapter that connects governed execution to a real external capability.
Receipt
Verifiable evidence describing the resulting governed execution.
Certification
Verification that the promoted execution path and its evidence satisfy the expected contract.
Default deny
Execution does not happen unless the required authority explicitly allows it.
Developer resources
Go deeper when you are ready.
Quickstart
Install the CLI and run your first project.
Start here
Orientation for the stable developer release.
CLI reference
Stable CLI usage and release information.
Versioning
Understand the stable channel and version policy.
Stable metadata
Machine-readable metadata for the current stable release.
Release metadata
Certified metadata for RBEK CLI 0.1.5.
AI Agent Governance
Public real-agent / real-Internet governed execution example.
Release & governance resources
Historical launch, adoption and feedback artifacts remain publicly available without dominating the getting-started experience.