AI Agent Governance

=== AI Agent Governance === Contributors: indraa Tags: ai, agent, governance, security, abilities-api Requires at least: 6.9 Tested up to: 7.0 Requires PHP: 8.0 Stable tag: 0.1.0 License: GPLv2 or later

Control, approve, and audit AI agent actions on your WordPress site.

== Description ==

AI Agent Governance sits between AI agents (ChatGPT, Claude, WPVibe, etc.) and your WordPress Abilities API. It provides:

• Kill Switch — instantly block all AI agent actions
• Policy Rules — wildcard pattern matching to allow, deny, or hold abilities
• Approval Queue — destructive actions require admin approval before execution
• Audit Log — complete trail of every AI agent attempt
• Blocked List — explicit blocklist for specific ability names or patterns

Looking for more? The Pro version adds Undo/Rollback, Email alerts, and Telegram notifications for pending approvals.

== Installation ==

1. Upload ai-agent-governance folder to /wp-content/plugins/
2. Activate through Plugins menu
3. Navigate to AI Governance in admin sidebar
4. Configure policies and blocked abilities

== Usage Guide ==

=== Quick Setup ===

1. Activate Plugin: Go to Plugins > Installed Plugins and activate AI Agent Governance.
2. Access Governance Dashboard: Click AI Governance in the main WP admin menu.

=== Core Features & Operations ===

• Emergency Control (Kill Switch)

  • Toggle the Emergency Kill Switch at top of page to ON to instantly block ALL AI agent actions across site.
  • Turn OFF to restore normal governance routing.

• Managing Policies & Rules

  • Click Add Rule under Policy Management.
  • Set Pattern (e.g. posts/*, users/delete, * for global default).
  • Select Action: Allow (pass through), Deny (block), or Hold (require admin review).
  • Rules evaluate top-down in display order.

• Approval Queue (Held Actions)

  • Actions marked Hold or flagged as destructive land in Pending Approvals.
  • Review action metadata and click Approve to execute or Reject to block.

• Blocked List

  • Add specific ability identifiers (e.g., core/update-settings) to Blocked Abilities for permanent block without rule evaluation.

• Audit Trail

  • Go to Audit Log tab to view timestamped history of all agent action attempts, parameters, decisions, and execution outcomes.

== Changelog ==

= 0.1.0 =

• Initial release