Abilities MCP · Helena site · Read-only session

Five calls.
One complete picture.

How Claude assembled the Priestess cohort 360 — from tag ID discovery through full journey fetch to rendered dashboard. Every step visible, every call documented.

Discovery

Call 1 of 5 readonly

mcp-adapter / discover-abilities

Searched twice — "contact" then "tag" — to confirm the CRM ability manifest and learn exact parameter schemas before making any data calls. Returned 41 abilities with input schemas, annotations, and tier info.

41 abilities confirmed schemas inspected annotations read

manifest confirmed → proceed to resolve

Resolve

Call 2 of 5 readonly

fluent-crm / list-tags

Searched "priestess" to resolve tag IDs before filtering contacts. Without this step, the contact query would require guessing numeric IDs. Returned two tags with subscriber counts.

ID 52 · AT_PRIESTESS_CH1_ONB · 24 subscribers ID 53 · AT_PRIESTESS_CH1_ORD · 20 subscribers

IDs resolved → parallel fetch begins

Fetch

Call 3 of 5 readonly

fluent-crm / list-contacts

Filter: tags "52,53" · per_page 100. Returns the full contact roster — IDs, names, emails, status, created dates — used as the selector for the next call.

24 contacts · all subscribed

Call 4a of 5 readonly

fluent-crm / get-cohort-journeys

Selector: contact_ids · all 24 IDs · include_notes: summary. Returns tags + tag_dates, purchases, email stats, automations, event keys + counts, note summaries. 10–15× cheaper than individual journey calls.

full journey per contact

journeys merged into context → patterns call

Analyse

Call 4b of 5 · second turn readonly

fluent-crm / get-cohort-patterns

Same 24 contact IDs · purchase_event_key: cart_order_paid. Returns aggregate purchase stats, time-to-first-purchase distribution, pre-purchase engagement signals ranked by frequency, tag progression paths, email engagement totals, and the full conversion funnel. ~2K tokens.

117.5k SEK total revenue 87.5% purchase rate 83.4% email open rate 10 engagement signals ranked

all data in context → render

Assemble

In-context · no further calls

Data assembled in Claude context

All three return payloads are merged in the conversation context. No additional ability calls needed to render either dashboard — the data is already complete.

list-contacts

id, name, email
status, created_at

get-cohort-journeys

tags + tag_dates
purchases, emails
automations, events
note summaries

get-cohort-patterns

purchase_stats
conversion funnel
engagement signals
tag_progression

rendered in-browser · no server calls

Render

Dashboard 1 · Turn 1

360 Cohort Dashboard

Roster table with filters, revenue bar chart, archetype coverage grid, entry path bars. Built from list-contacts + get-cohort-journeys data.

Dashboard 2 · Turn 2

Cohort Patterns

Conversion funnel, product revenue table, time-to-purchase distribution, engagement signals, email stats. Built from get-cohort-patterns data.

pipeline documented → this diagram

Document

Call 5 of 5 · this file readonly

Pipeline diagram · HTML artifact

The pipeline itself, diagrammed in WE brand language and exported as a standalone HTML file for use on site, in blog posts, and in communications. Dark and light mode. No external dependencies beyond Google Fonts.

dark mode light mode embeddable standalone HTML

Five calls.
Zero dashboards.
One conversation.

Wicked Evolutions · Abilities MCP · Helena site · Session 2026

Five calls. One complete picture.

Five calls.
One complete picture.