Desktop app + Claude plugins · built by operators, for small marketing teams

Turn Claude into your team's institutional brain.

A desktop app and a stack of Claude plugins that turn your marketing stack into a local, citation-disciplined wiki — then write the reports your team actually needs.

The Hive Mind vault, open in a markdown editor A macOS-style window showing a markdown editor. The left sidebar lists the vault directory tree: raw, sources, entities, concepts, syntheses. The right pane shows a sources page for a Q2 PMax wasted-spend audit, with YAML frontmatter naming the source file and linked entities, a Fraunces heading, body prose with bracketed wikilinks rendered in pressed-teal, a citations section listing the raw CSV with line numbers, and a footer chip reading "3 sources, 4 links, citation-disciplined". sources/google-ads-2026-q2-audit.md — hive-mind-vault VAULT ▾ hive-mind-vault ▸ raw/ ▾ sources/ shopify-orders-w24.md meta-ads-2026-q2.md google-ads-2026-q2-audit.md ga4-week-24.md slack-standup-2026-06-w3.md ▾ entities/ client-acme-corp.md campaign-q2-pmax.md ▾ concepts/ wasted-spend-by-geo.md ltv-by-cohort.md ▾ syntheses/ q2-paid-channel-review.md ▸ CLAUDE.md 247 PAGES · 1,284 LINKS LAST INGEST · 14 MIN AGO sources google-ads-2026-q2-audit.md --- source: google-ads-2026-q2.csv ingested: 2026-06-19 entities: [[client-acme-corp]], [[campaign-q2-pmax]] concepts: [[wasted-spend-by-geo]] --- Q2 PMax wasted-spend audit The Q2 PMax campaign for [[client-acme-corp]] spent $47,221 across April through June, of which roughly 23% went to out-of-market geos (rural ON, US East). See [[wasted-spend-by-geo]] for the per-region breakdown. Citations Spend → raw/google-ads-2026-06.csv L412–L498 Geo data → raw/google-ads-geo-2026-q2.csv Baseline → [[campaigns-2026-q1-pmax]] 3 SOURCES · 4 LINKS CITATION-DISCIPLINED
A markdown vault that lives on your computer. /ingest brings your stack in, /report-* writes the reports out.

Connects to Google Ads · Meta · Shopify · GA4 · Slack · Gmail · Linear · anything you can export

Small teams lose context every week.

Reaching for efficiency means losing specificity.

We live in a time of low-effort slop spit out by AI. The pressure is to move fast, but quality and detail suffer with AI models that know something about everything, but nothing about YOU.

Every report starts from a blank spreadsheet.

Last week's numbers live in five dashboards. Last month's context lives in someone's head. So every client update and channel report gets rebuilt from memory — and the story changes depending on who's telling it.

You want to do the work, but you're buried in the busywork.

Keeping a marketing team moving means a constant triage of clients, channels, approvals, logins, and threads no one has time to read twice.

Hive Mind fixes this by giving your team — and your AI — a single, cited, versioned source of truth that's constantly updated, automatically.

Four steps. Then it runs on its own.

  1. 01

    Install the Hive Mind desktop app and its Claude plugins. One-time setup.

  2. 02

    Connect your accounts — Google Ads, Meta, Shopify, GA4, Slack, Gmail, Linear. OAuth only; tokens stay in your computer's keychain. The app pulls each platform into raw/ on the schedule you set.

  3. 03

    Run /ingest. Claude classifies each new file, files it under sources/, entities/, concepts/, or syntheses/, and writes cited wiki pages — never modifying the raw source.

  4. 04

    Run the report skills: /report-internal (the internal .xlsx) and /report-client (the client .pptx), the Google Ads and Meta audits, and the daily /brief — deterministic files, every time.

A desktop app, a wiki, and a stack of plugins.

One knowledge base for your marketing team.

The wiki

Your team's institutional brain

CLAUDE.md
Operating contract. Defines the wiki maintainer's three operations and the hard rules below.
ingest
Brings a file from raw/ into the wiki, classifies it (source / entity / concept / synthesis), and links it to your existing pages.
query
Answers a question using cited pages. If a claim isn't grounded, it's flagged [uncertain].
lint
Sweeps the canon for orphans, stale claims, and broken [[wikilinks]]. Surfaces only — never auto-fixes.
Schema
sources/ (one page per ingested file) · entities/ (people, brands, products, places) · concepts/ (ideas, frameworks, methods) · syntheses/ (cross-cutting analyses).

The plugins

Pre-built skills that run on your data

operations-pack
The marketing-team OS. Onboard clients, ingest sources, enrich the wiki, and emit internal .xlsx + client .pptx reports. 13 skills, 10 commands.
clickt-google-ads
22 Google Ads skills — account audits, wasted-spend, bidding, structure, and optimization reports, all from a CSV export.
clickt-meta-ads
17 Meta Ads skills — creative, audience, structure, and Advantage+ audits straight from your exports.
calendar + linear
Pull your calendar and Linear work into the wiki, draft the daily standup, and distribute approved tasks. Draft-only — never auto-sends.
wppc-report
Wraps the wPPC analytics CLI for fast paid-media performance pulls.
contribution-scoreboard
Tracks who's actually moving the wiki forward — per-contributor output across the vault.

A desktop app · six Claude plugins · 50+ skills · one shared wiki vault. All authored by Clickt Digital Marketing Inc. Every report is deterministic — same inputs produce SHA-256-identical files across runs.

Rules that make it trustworthy.

Never fabricates.
Every claim cites a source, or it's flagged [uncertain].
Never modifies your raw sources.
raw/ is read-only forever.
Stops before destructive actions.
Asks before deleting, renaming, or restructuring.
Wikilinks over prose.
Knowledge connects.

Built for teams whose knowledge keeps walking out the door.

Small in-house marketing teams running paid, content, and SEO across more tools than people.

Boutique agencies juggling a dozen clients — and rebuilding the same report for each one every month.

Marketing leads who want to move fast, but refuse to ship AI slop that sounds pretty and means nothing.

Request early access.

Hive Mind is in private beta. We onboard a small number of marketing teams at a time — drop your email and we'll reach out the moment a spot opens.

  1. 01

    Request access above. Just your email — that's the whole form.

  2. 02

    If a spot is open, we email you an invitation to install Hive Mind.

  3. 03

    Install the desktop app and the Claude plugins. The setup wizard walks you through connecting your accounts.

  4. 04

    Connect your stack, run /ingest, and schedule your reports. You're live.

Proprietary. © Clickt Digital Marketing Inc. — all original work, no third-party content. Connectors: Google, Meta, Shopify, Slack, Linear, Granola. Hive Mind is the sole fetcher — tokens stay in your OS keychain and pulls proxy through our broker, but nothing persists on a server we control. Your data lives only in your local vault.