Domains

brain-mcp organizes your thinking into 25 cognitive domains — distinct areas of focus that each maintain their own state, open questions, and decision history.

What Are Domains?

A domain is a cognitive context — a distinct area where you think, decide, and create. When you switch from coding a React component to discussing Torah, you're switching domains. Each domain carries its own:

  • Thinking stage — exploring, crystallizing, refining, or executing
  • Open questions — unresolved threads within that domain
  • Decisions — choices you've made
  • Activity patterns — when and how you engage with this domain

The 25 Domains

ai-devLLM development, agents, MCP, embeddings
ai-strategyAI market, AGI, direction decisions
ai-imageImage generation, Midjourney, ComfyUI
automationCron jobs, scripts, system automation
backend-devAPIs, webhooks, server-side logic
business-strategyBusiness decisions, partnerships
careerJobs, resume, career direction
cognitive-architectureThesis, bottleneck, SEED, monotropism
data-engineeringPipelines, ETL, parquet, data cleaning
databaseSchema, migrations, Supabase, RLS
devopsDeployment, Docker, CI/CD, infrastructure
documentationDocs, README, wiki, writing
document-processingOCR, PDF, invoice parsing
financeBudget, taxes, accounting
frontend-devNext.js, React, Tailwind, UI
healthSleep, exercise, medication
mobile-deviOS, Android, mobile apps
optilabOptical, lens, glasses domain
personalFamily, kids, personal matters
prompt-engineeringSystem prompts, prompt design
pythonPython scripts, pandas, venv
torahParsha, halacha, Gemara, shiur
web-scrapingPlaywright, selectors, crawling
wotcWOTC tax credits, employer programs
generalCatch-all for unclassified conversations

How Domain Detection Works

Domains are assigned during the summary generation phase. When a conversation is summarized, the LLM classifies it into a primary domain based on conversation content. The classification considers:

  • Keywords and technical terms used in the conversation
  • The tools, frameworks, or concepts discussed
  • The type of questions asked and decisions made
  • Context from the conversation title and source
# See your domain distribution
mcporter call brain.brain_stats view="domains"

# Example output:
# Domain Map (9,979 conversations)
# ai-dev: 1,847 (18.5%) ████████▌ 🔥12 ⭐342
# frontend-dev: 1,203 (12.1%) ██████ ⭐198
# backend-dev: 987 (9.9%) █████ ⭐156
# ...

Thinking Stages

Each conversation within a domain is classified into one of four thinking stages:

🔍

Exploring

Asking questions, gathering information, considering options. You don't yet know what the answer is.

💎

Crystallizing

Patterns are emerging. You're forming opinions and making initial decisions. The shape of the solution is becoming clear.

🔧

Refining

The core decision is made; you're polishing, optimizing, or handling edge cases.

🚀

Executing

Building, shipping, deploying. The thinking is done; you're making it real.

Tools That Use Domains

Several brain-mcp tools accept a domain parameter to focus their output:

ToolWhat it does with domains
tunnel_state(domain)Reconstruct your save-state for a specific domain
context_recovery(domain)Full re-entry brief for a domain you're returning to
switching_cost(current, target)Cost of switching between two domains
tunnel_history(domain)Activity history within a domain
cognitive_patterns(domain)When you think best in a domain
dormant_contexts()Find abandoned domains with unresolved work
unfinished_threads(domain)Open questions within a domain

Explore your domains

Try mcporter call brain.tunnel_state domain="your-domain" to see where you are in any domain. Or use brain.dormant_contexts to find work you may have forgotten about.

PreviousData SourcesNextProsthetic Model