The Cortex command set to measure, report, and deeply understand every agent and client — from the monthly executive report to tracing a buyer back to the content that converted them. All run by talking to Claude.
Monthly executive report for the client: end-to-end conversion metrics + qualitative read, data-driven. Has a full internal version.
Monthly account status (internal): the health of the whole Ninjō client portfolio in a clean HTML, exportable to PDF.
Unified agent analysis: QA rubric, metrics-based evaluation, continuous-improvement loop, and prompt-fidelity audit.
Quick operational snapshot: executive summary with key metrics + 2-4 non-obvious findings cross-referencing several sources.
Finds where the funnel breaks (with data, not opinions) and builds scaling scenarios toward target sales.
Full per-contact traceability: lead origin, every content touchpoint, traffic temperature, and the "Winning Flow" common to buyers.
Cross-references buyers × content: matches each buyer to their conversation and the ad/story/content that triggered the conversion. Posts to Slack #data-agent.
The Ninjō Pixel: run the per-client performance metric, log feedback and proposals, and keep the hard measurement rules.
Generates 3 interactive dashboards per client (performance, period comparison, attribution) and deploys them to Cloudflare Pages.
Unified operational dashboard: business metrics (CSV+HTML), monitor mode with KPI alerts to Slack, and commit digest.
Multi-source pulse of your clients (Slack · git · DB · agent data): what's new since the last run, high-frequency.
Cross-agent fix mining: reads all fix-logs and changelogs and builds an evidence report of which techniques work.