codbrain
Connect & MCPSeller SettingsPrivate Wiki

Connect & Integration

codbrain is an MCP server. Your AI reads doctrine & procedures; your local machine executes with your credentials.

API keys (1 active)

Shown once upon creation. Store it in your own environment — we only keep a SHA-256 hash.

KeyCreatedLast usedAction
cb_live_9fA2••••••••2026-07-012026-07-27

Client Integration Setup

Select your client environment below for instant copyable configuration.

Run this single terminal command to pair Claude Code:

claude mcp add codbrain --transport http https://codbrain-two.vercel.app/api/mcp --header "Authorization: Bearer YOUR_KEY"

codbrain never asks for your credentials. Your Meta, Vercel and Apify keys stay on your machine. The brain returns doctrine and procedures; your AI executes locally.

Interactive Tool Playground

Simulate MCP tool responses directly in your browser without terminal calls.

Live Sandbox

Select Tool

Arguments (JSON)
Response Output
// Select a tool and click 'Execute Tool' to test.

Tool Engine Specifications

15 specialized tools available to any connected AI agent.

15 Tools Active
codbrain_start

Read first, once per task. Returns execution rules, route for task, and writes CODBRAIN.md file.

check_setup

What needs wiring before a task is executable — credentials to ask for vs local tools to install.

get_profile

Reads your seller settings & real fee stack so your AI works from factual unit economics.

brain_ask

A composed answer from doctrine, with sources, evidence levels, and market scope checks.

get_procedure

Step-by-step checklists with hard gates — build store, make static/animated ads, launch, scale.

check_economics

Computes break-even CPL from seller profile & reported rates. Refuses rather than guessing.

list_templates

Store templates (gadget, cosmetics) you can clone instead of building funnels from scratch.

get_template

Fetch signed template link — AI downloads & re-brands locally on your machine with your keys.

brain_index

Catalogue of your private wiki by category, keeping track of what was already worked out.

brain_note

Write conclusions to your private seller wiki so work is never re-derived next session.

brain_recall

Read private wiki pages back. Fully isolated via Supabase RLS.

brain_log

Running audit of actions tried and costs incurred to prevent duplicate failures.

get_benchmark

Network medians for your market & niche. Locked until you contribute.

report_outcome

Report test outcomes with strict schema validation & automatic secret scanning.

contribute_lesson

Send back qualitative insights — why an angle failed or how a bug was resolved.