The control plane for WordPress agents

AI agents can act in WordPress. You decide how far.

Put one explicit governance boundary between every managed AI agent and your WordPress Abilities—permission, approval, evidence, and emergency control in one local-first system.

Download Free See the control path
Managed request path
Production · Policy active
01 · REQUEST
Content OperatorManaged agent · Production
WordPress Abilityacme/update-post
Scope1 published post
RuleFence
02 · DECISIONApproval requiredMedium risk
Identity verified Capability allowed Human review
03 · WORDPRESS
Registered AbilityNative authorization preserved
Execution stateLocked pending approval
EvidenceAudit record prepared
Default deny active
Native capability remains the ceiling
Signed evidence for every decision
Emergency stop always available
  • Local-first core
  • No required AI API key
  • No per-call fees

The control loop

One governed path from connection to evidence.

Every managed request moves through the same explicit boundary—identity, permission, policy, approval, execution, evidence, and supported recovery.

01

Identify

Give every client its own agent identity, connection, state, and environment.

02

Control

Decide Ability by Ability: Allow, Require Approval, or Block.

03

Approve

Bind a human decision to the exact agent, Ability, input, and expiry.

04

Verify

Record the decision, result, outcome evidence, and audit integrity.

05

Recover

Restore supported before-state when a recovery provider can do so honestly.

Complete governance

Connection is only the beginning.

RuleFence controls what happens after an agent connects—without rebuilding WordPress authorization or trapping you in a proprietary tool catalog.

Explore the product
ABILITY EXPLORER

Know what agents could potentially do.

Discover registered Abilities, providers, schemas, exposure, action type, risk, and current agent access before granting anything.

Explore Abilities
PERMISSION MATRIX

Grant the smallest useful permission.

Configure explicit decisions per agent. Anything unconfigured stays blocked.

APPROVAL CENTER

Review the exact request—not a vague permission.

See the agent, Ability, risk, expected change, input preview, and expiry before deciding.

GOVERNED OPERATIONS

Turn controlled actions into repeatable operations.

Pro adds contextual policy, before/after previews, supported recovery, workflows, schedules, notifications, and detailed run evidence.

See Pro features
SIGNED ACTIVITY

Keep an accountable local record.

Read activity in human language and verify that retained events remain cryptographically linked and intact.

Security model

Your site remains the control boundary.

Agent identities, permissions, approvals, policies, sessions, and local activity stay in your WordPress database by default. Native WordPress authorization always remains the maximum authority.

  • Dedicated least-privilege connections
  • Exact, single-use approvals
  • Fail-closed authorization and evidence
  • Signed audit integrity verification
  • Immediate pause, restrict, and revoke
Read the security model
Runtime Verification
Verified
100/100
READINESS SCOREControl path healthy

All required governance checks passed.

WordPress Ability lifecyclePassed
Default-deny enforcementPassed
One-time approval behaviorPassed
Signed audit chainPassed
Credential redactionPassed
Verification does not execute write-capable Abilities.

Built around WordPress Abilities

A governance layer—not another connector.

Use the connection your client needs. Keep one durable control model for identity, permissions, approvals, evidence, recovery, and fleet governance.

01

Abilities first

Govern functionality registered by WordPress and plugins instead of rebuilding a closed tool catalog.

02

Protocol independent

MCP is a supported connection option—not the permanent identity of the product.

03

Local first

Core governance runs on your WordPress site without forcing a vendor cloud into the execution path.

04

Narrower, never broader

RuleFence can reduce agent authority; it cannot override a WordPress capability denial.

Simple annual pricing

Basic safety stays Free.

No per-action fees. No AI credits. No required cloud account for core governance. Upgrade when your agents become an operational system.

Free

For getting started with managed AI access, on as many sites as you run.

$0forever · unlimited sites
  • Every site you run
  • Unlimited active agents
  • Ability Explorer
  • Allow, Approval, and Block
  • One-time approvals
  • Recent signed activity
  • Sessions and emergency controls
  • Runtime Verification
Download Free
Recommended
Pro

For a site that depends on AI agents for real work.

$79/ year · 1 site
Early-adopter rate. $99 after 31 December 2026.
  • Everything in Free
  • Contextual risk and policies
  • Rich previews and verification
  • Supported snapshots and undo
  • Governed workflows and schedules
  • Extended audit and export
  • Priority email support
Get Pro
Pro, 5 sites

For a small portfolio held to one boundary.

$199/ year · 5 sites
Early-adopter rate. $249 after 31 December 2026.
  • Everything in Pro
  • Five production sites
  • A staging site alongside each one
  • One licence key across them all
  • Priority email support
Get Pro 5
Agency

For agencies applying one governance standard across client sites.

$399/ year · 25 sites
  • Everything in Pro
  • Fleet inventory and site groups
  • Fleet readiness visibility
  • Central approval queue
  • Policy deployment and drift
  • Fleet emergency restriction
  • Target-site signed evidence
  • Priority onboarding and support
Get Agency

All prices in USD before applicable tax. Staging and local development copies of a licensed production site do not consume an extra activation.

Designed for accountable AI operations

One boundary. Different teams.

Site owners

Start with read access, approve meaningful changes, and keep an emergency stop close.

Learn more

Content & SEO

Let agents draft and update while publishing, deletion, and bulk scope stay controlled.

Learn more

WooCommerce

Put price, stock, product, and order consequences behind compatible policies and approval.

Learn more

Developers

Give maintenance agents useful access without turning production into a playground.

Learn more

Agencies

Apply governance standards across sites while preserving every client’s own audit boundary.

Learn more

Questions, answered

Clear boundaries from the start.

RuleFence is built to make authority, evidence, limitations, and recovery coverage visible—not implied.

Read all documentation
Is RuleFence an MCP server?+

No. It is an agent-governance layer built around WordPress Abilities. The official MCP Adapter is an optional connection path.

Does it replace WordPress roles and capabilities?+

No. WordPress authorization remains the maximum authority. RuleFence adds narrower agent-specific permissions and controls.

Does the plugin include an AI model?+

No. Bring your own compatible AI client or agent. RuleFence controls WordPress access and does not resell model tokens.

Can every action be undone?+

No. Undo is offered only when a registered provider can capture and restore the relevant state. External side effects may not be reversible.

Is a cloud account required?+

No for core local governance. Any optional hosted service is separate, clearly disclosed, and deliberately enabled.

Start with a boundary

Give your next AI agent control before you give it access.

Start with visibility. Grant one useful permission. Keep everything else blocked.

Download Free Read getting started