ABAgentBlackBoxv2.0
ProductDocsPricingSecurityPilot program
Sign inGet started
ProductDocsPricingSecurityPilot programSign in

Product

Visibility, safety controls, human oversight and evidence for your AI agents

Send your agents' actions through AgentBlackBox, or connect the apps they work in. Rules decide what may happen, a person approves the risky steps, and every step joins a record you can verify and export.

AgentBlackBox sees and controls only what your agents send through it or through tools you connect. It does not monitor other AI tools.

Create a free workspaceSee the product tour

In the product

Screenshots of AgentBlackBox with example data.

The Approvals page with a held 4,800 payment, its risk and the agent that asked, with Approve and Deny buttons
A payment waits for a person. The approver sees the action, the amount, the risk and the agent.
A task's timeline with three hash-chained events and a valid hash chain
Every step joins a hash-chained record; the chain verifies.
The Rules page with rule templates and the workspace's rules
Rules in plain terms, with templates and version history.

What each label means

  • Available On for every workspace today.
  • Set up per workspace Available; your workspace sets it up.
  • On request Built; we switch it on for a pilot workspace.
  • Not switched on yet Built, not switched on in our service yet.
  • Planned Not built yet.

See what your agents do

A record of the activity your agents report or route through AgentBlackBox, and of what happens in the apps you connect.

Activity record per task

Available

Agents send what they observed, decided and did, through the API or the SDK. Each task gets its own timeline.

Connected apps

Available

Slack, Gmail, Google Drive, GitHub and Notion, connected with their own sign-in. Activity from them joins the record.

Agent registry

Available

Each agent has a name, a risk class and the tools it may use. An action outside its tools is refused; an agent can be switched off.

Reports

Available

Decisions, approvals and blocked actions over 7, 30 or 90 days, with a signed export.

Decide what they may do

Rules check every action your agents send through AgentBlackBox, before it runs.

Rules

Available

Allow, ask a person first, or block, by action type, risk, amount, recipient, tool or content. Rules can also change or redact an action. Built-in rules already block deletions and ask before money moves or data leaves your organization.

Watch-only first

Available

New workspaces start in watch-only mode: rules decide and record, nothing is blocked. Switch to enforce when the decisions look right.

Risk assessment

Available

AgentBlackBox sets the risk itself: sending outside your organization or moving money raises it, and an agent can never lower it.

Execution controls

Available

Actions go through three gateways: agent decisions, connected-app actions and custom tools. A held action runs only after it is approved. App credentials stay encrypted with AgentBlackBox; the agent never holds them.

Rule history and testing

Available

Every rule change is a new version you can roll back. "Test a rule" shows the decision for a sample action without recording it.

Rate limits

Available

Each agent key has limits on the gateways. If a limit cannot be checked, the action is refused rather than run unchecked.

Keep a person in the loop

Risky actions wait for a human decision, recorded with the action and its risk.

Human approvals

Available

In a team, the approver must be someone other than the person or agent that asked. Approver groups decide who may approve what.

Personal workspaces

Available

Working alone? You confirm your own risky requests by entering your password again, and the record notes it.

Approval safeguards

Available

A request expires after 72 hours without a decision, and approving checks the action against the rules in force at that moment.

Workflows

Available

Multi-step workflows with an approval step, run on demand.

Scheduled workflows and reminders

Not switched on yet

Workflows on a schedule, reminders for waiting approvals and escalation to a group's backup approver.

Keep evidence you can check

A tamper-evident record, and exports for auditors and clients.

Tamper-evident records

Available

Every recorded event is hashed with SHA-256 and chained to the one before it. Verifying shows whether a recorded event was changed or removed from the middle of the chain.

Checkpoints

Available

Signed checkpoints seal a range of events, so removing the most recent ones is detected too. Admins create them; automatic checkpoints need scheduled jobs.

Signed exports and reports

Available

A JSON export signed by AgentBlackBox (HMAC-SHA256) and a printable audit report for every task; owners can export the whole workspace. Checking a signature needs AgentBlackBox's key.

Audit log

Available

Who changed which setting, rule, key or member, and when.

Public anchoring

Not switched on yet

Anchoring checkpoint hashes to a public timestamp service (OpenTimestamps).

Signatures anyone can check

Planned

Public-key signatures, so a third party can check an export without our key.

For developers

Add a policy check and an approval step to your agent with a few API calls.

REST API

Available

Record events, evaluate rules, request and poll approvals, and send actions through the gateways. Every answer carries a request id.

API keys

Available

Personal or service keys, optionally bound to one agent and given an expiry. Shown once and stored only as hashes.

Custom tools

Available

Register your own HTTP endpoints and MCP servers; agents call them through the same rules. Only public https addresses are accepted.

Signed webhooks

Available

Inbound webhooks are signed with a timestamp and refused when replayed.

TypeScript SDK

Available

A client for every endpoint, including the modify, redact and retry answers. Available in the repository; not published as a package yet.

For teams and businesses

Roles, company sign-in and the controls a security team asks for.

Roles and members

Available

Owner, admin and member. Owners manage owners and role settings; admins can deactivate members; a workspace always keeps an active owner.

Single sign-on and SCIM

Set up per workspace

Microsoft Entra sign-in with proof that your organization owns its email domain, and automatic member provisioning.

Legal hold and privacy requests

Available

Keep all records under legal hold; export or erase a person's data.

SIEM streaming and Slack notifications

On request

Stream the audit log to Splunk, Datadog or a webhook, and notify Slack channels.

Your account

Sign-in and recovery.

Secure sign-in

Available

Passwords stored as bcrypt hashes; password sign-in pauses after 10 wrong attempts; changing your password or "Sign out everywhere" ends every session.

Password reset and email verification by email

Not switched on yet

Single-use links that expire. Until email sending is switched on, support helps you back in.

Two-factor sign-in

Planned

A second factor for password accounts.

Paid plans

Not switched on yet

Personal, Pro and Business subscriptions through Stripe (coming soon, billed monthly). Every feature works on the free plan today.

What AgentBlackBox does not do

  • It does not watch AI tools you have not connected, or actions an agent takes with its own credentials outside AgentBlackBox.
  • It checks the actions sent through it; it cannot stop an action that never reaches it.
  • It holds no SOC 2, ISO 27001 or similar certification, and using it does not by itself make you compliant with any regulation.
How it is securedDeveloper quickstartJoin the pilot program

AgentBlackBox

AgentBlackBox checks what your AI agents try to do against your rules, asks a person before risky actions, and keeps a tamper-evident record of what happened.

AgentBlackBox sees and controls only what your agents send through it or through tools you connect. It does not monitor other AI tools.

Product

  • Overview
  • Product
  • Product tour
  • Pilot program
  • Pricing
  • Documentation

Trust

  • Security
  • Report a vulnerability
  • Privacy
  • Terms
  • Subprocessors

Help

  • Help center
  • Support
  • Contact

© 2026 AgentBlackBox. All rights reserved.