Skip to main content
Glama

Command Center

Your company's purchased productivity tools, inventoried once — then usable as a second brain by any AI assistant. One repo, one source of truth, zero per-seat AI tax.

The idea

Every company pays for a sprawl of tools — email providers, campaign platforms, team chat — and nobody holds the full picture. Command Center fixes that in two layers:

  1. Inventory — inventory.yaml records every purchased tool: plan, seats, cost, owner, renewal date. discovery/ scripts detect what's actually in use (MX/DNS checks, expense keyword lists) so the inventory stays honest.

  2. Second brain — brain/server.py is an MCP server. Claude Desktop, ChatGPT, Cursor and friends connect to it and can answer "which tools do we pay for that do the same thing?" or "what renews next quarter?" straight from your inventory, as typed context packets. Read-only by design.

Related MCP server: toolscope

Repo map

Path

What it is

catalog/tool-catalog.md

Curated catalog: 19 email/campaign/comms tools with verified API auth, base URLs, and passive discovery signals

discovery/mx_discover.py

Passive email-provider detection via MX + SPF records (no credentials)

inventory.example.yaml

The inventory schema — copy to inventory.yaml, fill in your stack

brain/server.py

MCP server: list_stack, get_tool_context, find_overlap, stack_summary

requirements.txt

fastmcp, dnspython, pyyaml

Quickstart

python3 -m venv .venv && .venv/bin/pip install -r requirements.txt

# 1. Detect a company's email provider (passive, no credentials)
.venv/bin/python discovery/mx_discover.py zoho.com

# 2. Record your stack
cp inventory.example.yaml inventory.yaml   # then edit it

# 3. Start the second brain (stdio MCP server)
.venv/bin/python brain/server.py

Register brain/server.py as an MCP server in Claude Desktop (Settings → Connectors → Add custom connector) or ChatGPT (Settings → Apps / MCP). Both speak MCP natively — one server, both assistants.

Why this, why now

SaaS inventory tools (Zylo, Torii, Vertice) are enterprise-quote-only with $50K+ floors. AI search tools (Glean) can't act; agent tools (Moveworks) cost six figures. And cloud second brains can vanish overnight — Limitless was acquired by Meta and shut down in Dec 2025. This is the MIT-licensed, self-hosted, MCP-native alternative: your context layer, owned by you. Tokens are rented. Context is owned.

Roadmap

  • OAuth connector guides per catalog tool (least-privilege scopes)

  • Expense CSV import → auto-suggest inventory entries

  • Renewal reminders (scheduled brief)

  • Tiny web dashboard over inventory.yaml

Maintenance

ActivityMaintained
ResponsivenessNo issues

Related MCP Connectors

Related MCP Servers

  • A
    license
    Not graded
    quality
    C
    maintenance
    Enables AI agents to query a local SQLite catalog of installed macOS GUI apps and CLI tools, searching by capability, retrieving tool details, and checking machine statistics over MCP.
    MIT
  • A
    license
    Not graded
    quality
    B
    maintenance
    Enables AI assistants to read current pantry inventory and low-stock shopping needs, export a shopping queue, and perform idempotent inventory adjustments and catalog link management through remote MCP tools.
    MIT
  • A
    license
    Not graded
    quality
    B
    maintenance
    Enables MCP-compatible agents such as Claude and Cursor to retrieve permission-aware, structured context from local or hosted data stores. Exposes tools for semantic search, entity and relationship lookup, memory access, context pack compilation, source attribution, and retrieval explanation.
    1
    Apache 2.0