coppermind-cmo 0.4.25

AI-powered memory for fractional CMOs — never forget a client detail, never leak data between clients

Coppermind CMO

AI-powered memory for fractional CMOs. Never forget a client detail, never leak data between clients, always prep well for meetings.

Coppermind CMO is an MCP server that gives Claude persistent, client-scoped memory across conversations. Store meeting notes, brand voice, campaign outcomes, and decisions — then search, prep, and recall instantly.

Installation

pip install coppermind-cmo

Setup

Add to your Claude Code settings (~/.claude/settings.json):

{
  "mcpServers": {
    "coppermind-cmo": {
      "command": "python",
      "args": ["-m", "coppermind_cmo"],
      "env": {
        "COPPERMIND_API_KEY": "your-key-here"
      }
    }
  }
}

Note: Use python3 instead of python on Mac/Linux if python is not aliased to Python 3.

That's it. The API key handles everything — database access, embeddings, and extraction services are all managed for you.

What it does

48 tools across 11 categories:

Category	Tools	Purpose
Core (15)	switch_client, list_minds, configure_mind, store_memory, search_memory, quick_note, hide_memory, get_brand_voice, set_brand_voice, prep_meeting, get_last_brief, get_campaign_history, list_documents, save_output, done	Client management, memory storage/search, brand DNA, meeting prep
Export (1)	export_mind	GDPR-ready full mind export as JSON or Markdown — memories, brand voice, rocks, campaigns, stakeholders
Daily Loop (6)	get_weekly_summary, get_cross_client_summary, get_meeting_context, log_meeting_event, mark_followup_sent, ensure_personal_mind	Morning briefings, cross-client dashboard, meeting follow-ups
Auto-Ingest (2)	auto_ingest_transcript, find_uningested_meetings	Automatic meeting transcript ingestion from connected sources (Granola, Otter, Fireflies, Fathom, Read.ai)
Email Ingestion (2)	ingest_emails, email_status	Source-agnostic email ingestion — works with any email MCP (Gmail, Outlook, IMAP)
Synthesis (1)	synthesize_document	Deep second-pass analysis of ingested documents — patterns, trends, contradictions, and insights
Knowledge Ingestion (1)	seed_cmo_knowledge	Bulk-ingest CMO professional knowledge from a local folder with LLM-assisted classification and client routing
Knowledge Health (1)	get_knowledge_health	Score 8 knowledge dimensions per client mind — find gaps and get actionable suggestions
Utility (1)	update_coppermind	Self-update to latest PyPI version via uv or pip
EOS (13)	get_rocks, manage_rock, get_issues, manage_issue, get_agenda, manage_agenda, get_current_sprint, get_sprint_plan, import_sprint_plan, update_sprint_status, quarterly_pacing, start_quarter, prep_l10	Rock tracking, sprint planning, L10 meeting prep, quarterly pacing
Team (7)	add_team_member, remove_team_member, list_team, list_pending, promote_pending, dismiss_pending, rollback_pending	Share client minds with VAs/team members with approval workflows

Key properties:

• Client isolation — every query is scoped to the active client. Zero cross-client data leakage.
• Brand DNA — store and retrieve brand voice, positioning, stakeholders, visual identity, anti-patterns, and platform-specific content guidance per client.
• Meeting prep — topic-aware retrieval across memories, commitments, and brand DNA, returned as a structured brief. Interactive follow-up questions supported.
• Team sharing — grant VAs and junior marketers filtered access to client minds. Viewer-created memories require CMO approval. Sensitive content auto-detected and hidden.
• Client-colored output — brand colors render in tool responses and terminal output for visual client context.
• Resilient — all tool calls are wrapped with structured error handling. Gateway calls retry with exponential backoff. Graceful degradation when services are temporarily unavailable.
• EOS-native — full EOS methodology support: rocks, sprints, L10 prep, quarterly pacing, and quarter transitions.

Daily workflow

# Before a call — get briefed
"Switch to Acme Corp and prep me for our meeting about Q2 planning"
→ Switched to Acme Corp. 47 memories, last meeting Mar 20, brand voice: set (850 words), EOS: Q2 2026 active.

# Interactive meeting prep — ask follow-up questions
"Tell me more about their Q1 campaign results"
"What's the history with Sarah Chen?"
"Compare to last meeting"

# During/after a call — capture what happened
"Remember that Acme decided to pause LinkedIn ads through Q2"
"Quick note: Sarah mentioned the April budget review might affect Q3"

# Switching clients — change context
"Switch to Bluebell"
"What do we know about Bluebell's campaign performance?"

# Quarter transitions
"Transition Acme to Q2 — carry forward the partner portal rock"

# Team sharing
"Add my VA to Acme Corp as a viewer"
"Review pending memories from my team"

# Seed your knowledge base
"Seed my knowledge from ~/Documents/CMO-Playbooks"
→ Scans, classifies, and routes files to personal mind or matched client minds

# Generate on-brand content
"/cmo-write a LinkedIn post about our Q2 results"
"/cmo-write draft a follow-up email to Sarah after today's meeting"
→ Multi-expert content panel with brand voice enforcement and style learning

Optional Extras

For local embedding or related infrastructure dependencies:

pip install coppermind-cmo[self-hosted]

See full documentation for current setup details.

Support

Contact Ben directly at [e-mail].