This project is not covered by Drupal’s security advisory policy.
AI Watchdog gives your Drupal site a self-healing nervous system. Autonomous AI agents continuously monitor, diagnose, and fix common site problems — broken links, accessibility failures, security drift, and content quality issues — so your team can focus on building, not firefighting.
This is the first open-source CMS module to bring self-healing infrastructure concepts into the Drupal ecosystem. Every action is logged, auditable, and reversible. Nothing happens without your explicit permission level.
Features
AI Watchdog ships five specialised monitoring agents, each targeting a different dimension of site health:
- Link Integrity Agent — Crawls your site for broken internal and external links. Can automatically create redirects or notify content authors, depending on your configured trust level.
- Accessibility Agent — Audits published content against WCAG 2.1 AA. Goes beyond rule-based checkers — uses AI to evaluate whether alt text is actually meaningful in context, not just present.
- Security Posture Agent — Monitors for known CVEs, insecure file permissions, configuration drift, and data sovereignty violations. Aligned with ISM and Essential Eight frameworks.
- Content Quality Agent — Checks reading level against plain language standards (useful for government sites), detects orphaned content, and flags potential contradictions across your content.
- Performance Agent — Tracks response times, cache hit rates, and slow queries. Can automatically enable CSS/JS aggregation and other quick wins.
- Analytics Dashboard — A visual overview of your site's health history. Charts show finding trends, agent activity, and auto-fix success rates over time — giving teams and stakeholders a clear picture of site health at a glance. (Coming soon)
Four trust levels give you full control over autonomy:
observe— Agents report only. Zero changes made.suggest— Agents create draft fixes for human review.auto_fix_low— Auto-fix low-risk issues (broken links, missing alt text).auto_fix_all— Full autonomy. Requires explicit opt-in.
Every autonomous action creates an AgentAction entity with full before/after state, so everything is auditable and rollback-ready.
Post-Installation
After enabling the module:
- Go to Admin → Configuration → AI Watchdog to configure global settings and per-agent trust levels.
- Choose which agents to enable and set each agent's trust level independently (e.g. observe for Security, auto_fix_low for Link Integrity).
- Agents run automatically on cron, or you can process them via Symfony Messenger queues for background execution on high-traffic sites.
- View findings and agent run history at Admin → Reports → AI Watchdog Dashboard — an interactive dashboard with charts showing agent run history, finding trends over time, and fix success rates across all agents. (Coming soon)
- Any auto-applied fixes can be rolled back individually from the AgentAction log.
Additional Requirements
- Drupal AI ^1.3 — Required. Provides the LLM provider abstraction layer. Works with any configured AI provider: OpenAI, Anthropic, Mistral, or a local Ollama instance for air-gapped deployments.
- AI Agents — Required. Provides the agent framework and tool calling infrastructure.
- Redirect — Required by the Link Integrity Agent for automatic redirect creation.
- Key — Recommended for secure API key storage when using cloud LLM providers.
- PHP 8.2+ and Drupal 10.4+ or Drupal 11.x
Recommended modules/libraries
- Ollama — Run a local LLM for fully air-gapped deployments. No data leaves your infrastructure. Ideal for government and defence environments.
- Symfony Messenger (included in Drupal core) — For background queue processing on high-traffic sites, avoiding cron timeouts.
Similar projects
The Drupal AI ecosystem provides building blocks for AI features. AI Watchdog is the first module built on top of that ecosystem specifically for autonomous site operations — it is not a content generation tool or a chatbot. If you want to add AI to your editorial workflow, look at AI Interpolator or ECA + AI. AI Watchdog's focus is keeping your site healthy without human intervention.
Supporting this Module
AI Watchdog is maintained by Amjad Pathan and proudly supported by Doghouse Agency.
Contributions, issue reports, and merge requests are welcome on the project page.
If this module saves your team time, consider sponsoring Drupal development or crediting your organisation on the project page.
Community Documentation
Full architecture documentation is available in the module repository at docs/ARCHITECTURE.md.
A DrupalPod demo and video walkthrough are planned for the Phase 2 release.
Project information
- Project categories: Artificial Intelligence (AI)
- Ecosystem: Security
- Created by amjad1233 on , updated
This project is not covered by the security advisory policy.
Use at your own risk! It may have publicly disclosed vulnerabilities.

