Capabilities

Features

Everything the Agent can do — from ingesting provider signals to executing Claude Code fixes with your approval.

Data & signals

  • Multi-provider signal ingestion

    Pulls clicks, impressions, CTR, and position from Google Search Console and Bing Webmaster Tools on a daily schedule. Keyword volume and SERP feature data from DataForSEO.

  • AI referral detection

    Identifies traffic originating from ChatGPT, Perplexity, Gemini, Copilot, and other AI surfaces using referrer and UTM pattern matching against the WebmasterID Core signal graph.

  • Core Web Vitals tracking

    CWV metrics from Google Search Console (field data) aggregated per URL. Regressions trigger work package creation automatically.

  • Deploy correlation

    GitHub commit and Vercel deploy events are correlated with ranking and performance shifts — surfacing regressions introduced by specific deployments.

Analysis & recommendations

  • Rule-based audit engine

    Deterministic rule set over fused signals. No LLM in the audit path — findings are reproducible: same inputs always produce the same recommendations.

  • Ranking drop detection

    Statistical comparison of position trends over configurable windows. Drops beyond threshold generate high-priority work packages with the affected queries and URLs.

  • Structured data gap analysis

    Detects missing or invalid JSON-LD on key landing pages. Surfaces opportunities for Article, FAQPage, Product, BreadcrumbList, and SoftwareApplication schemas.

  • Prioritised work queue

    Findings sorted by estimated impact (traffic risk × fix complexity). Each work item includes the triggering signal, affected URLs, severity, and suggested remediation.

Execution & approval

  • Claude Code MCP bridge

    Exposes Agent context to the Claude Code CLI via the Model Context Protocol. Claude can read signals, draft content, write code fixes, and apply changes — all locally.

  • Human-approval gate

    Every external action — commit, deploy, outreach — requires explicit operator sign-off. The approval gate is non-bypassable by the Agent or Claude Code.

  • Full audit trail

    Every sync run, provider fetch, recommendation, and approval event is logged with timestamp, actor, and outcome. Exportable for compliance review.

Frequently asked questions

Is the rule-based engine customisable?

The core rule set ships with WebmasterID Agent. Custom rules per workspace are on the M7–M12 roadmap. Today, threshold configuration (e.g. ranking drop sensitivity) is available per site.

How does AI referral detection work without JavaScript on the AI surface?

AI surfaces like ChatGPT send a recognisable referrer header when they link out. WebmasterID Core pattern-matches these against a maintained registry of known AI referral domains and UTM patterns.

Can I disable specific features per site?

Yes. Each provider data source can be enabled or disabled per site from the Provider data sources panel on the site detail page.