About bolor-brain-mcp
bolor-brain-mcp is an MCP server that adds structured reasoning, persistent memory, and learning from experience to Claude Code. It combines several reasoning engines with local JSON persistence and does not execute tasks.
Key features
- 01Select a reasoning approach automatically
- 02Use symbolic reasoning and knowledge graphs
- 03Reason from cases and hypotheses
- 04Store and retrieve cases and facts
- 05Delete memory entries by ID
- 06Persist state in JSON files
Use cases
- 01Analyze complex problems
- 02Investigate bugs through hypotheses
- 03Compare technical options
- 04Save solutions for later retrieval
MCP support
The MCP server provides Claude Code with reasoning, memory, and statistics tools; it does not perform file operations or other task execution.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
npx -y bolor-brain-mcpBRAIN_USE_SQLITEEnable SQLite storage (default: true)BRAIN_LOG_LEVELLog level (default: INFO)Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEBolor Brain MCP
Pure intelligence for Claude Code. Reasoning, memory, and learning -- nothing else.
What Is This?
Bolor Brain is an MCP server that gives Claude Code a brain: structured reasoning, persistent memory, and learning from experience.
User --> Claude Code (Gateway + Executor)
|
+----+----+
Bolor Brain NSAF
(MCP) (MCP)
THINK EVOLVE
- Claude Code = Gateway + tool execution + sessions + permissions
- Bolor Brain = Reasoning engines + memory + learning + persistence
- NSAF = Strategy evolution + self-improvement (separate MCP server)
Bolor Brain does NOT execute anything. No file ops, no scheduling, no autonomous loop. Claude Code already does all of that. Bolor Brain only thinks.
Quick Start
1. Install
git clone https://github.com/photoxpedia/bolor-brain-mcp.git
cd bolor-brain-mcp
pip install -e .
2. Configure Claude Code
Add to ~/.claude/mcp-config.json:
{
"mcpServers": {
"bolor-brain": {
"command": "python",
"args": ["-m", "mcp_server"],
"cwd": "/path/to/bolor-brain-mcp"
}
}
}
3. Use
/reason Why is Python popular for data science?
/debug API returns 500 errors under load
/decide PostgreSQL or MongoDB for our app?
/learn-from We fixed the memory leak by increasing connection pool
MCP Tools (11)
Reasoning (6)
| Tool | What It Does |
|---|---|
reason_hybrid | Auto-selects best reasoning approach for any query |
reason_symbolic | Forward/backward chaining with facts and rules |
reason_knowledge_graph | Graph traversal, path finding, relationship exploration |
reason_case_based | Find similar past problems and their solutions |
reason_hypothesis | Generate and test hypotheses from observations |
reason_analogical | Cross-domain pattern transfer (atom ~ solar system) |
Memory (4)
| Tool | What It Does |
|---|---|
remember | Store a case, fact, node, or edge |
recall | Retrieve matching cases or facts |
learn | Store problem/solution/outcome (shortcut for remember) |
forget | Delete a case or fact by ID |
Utility (1)
| Tool | What It Does |
|---|---|
brain_stats | Cases, facts, nodes, edges count |
Skills (6)
| Skill | When To Use |
|---|---|
/reason | Deep analysis of any complex problem |
/debug | Systematic bug hunting with hypothesis testing |
/decide | Evidence-based technical decisions |
/learn-from | Store experiences for future use |
/nsaf | NSAF evolution integration (requires NSAF MCP) |
/orchestrate | Meta-orchestration combining Bolor Brain + NSAF |
Persistence
Brain state persists to ~/.bolor-brain/ as JSON:
~/.bolor-brain/
cases.json # Problem -> solution -> outcome
facts.json # Symbolic reasoning facts
knowledge.json # Knowledge graph (nodes + edges)
Knowledge compounds over time. Solve a bug once, recall the solution instantly next time.
With NSAF
Add NSAF to get evolution capabilities:
{
"mcpServers": {
"bolor-brain": {
"command": "python",
"args": ["-m", "mcp_server"],
"cwd": "/path/to/bolor-brain-mcp"
},
"nsaf": {
"command": "python3",
"args": ["nsaf_mcp_server.py"],
"cwd": "/path/to/nsaf",
"env": { "PYTHONPATH": "/path/to/nsaf" }
}
}
}
Together: Bolor Brain reasons about WHAT to do. NSAF evolves HOW to do it better. Claude Code executes.
See skills/nsaf.md and skills/orchestrate.md for combined workflows.
Testing
pytest tests/ -v
# 376 tests
Project Structure
mcp_server.py # MCP server (11 tools)
persistence.py # JSON persistence to ~/.bolor-brain/
modules/
config.py # Configuration
reasoning_engines/
symbolic_reasoner.py # Forward/backward chaining
knowledge_graph.py # Graph-based knowledge
case_based_reasoner.py # 4R cycle (retrieve, reuse, revise, retain)
hypothesis_engine.py # Hypothesis generation and testing
analogical_reasoner.py # Cross-domain pattern transfer
hybrid_reasoner.py # Orchestrates all 5 engines
skills/ # Claude Code skills
reason.md, debug.md, decide.md, learn-from.md, nsaf.md, orchestrate.md
tests/ # 376 tests
AGENT_GUARDRAILS.md # Production safety guidelines
Author
Bolorerdene Bundgaa
- Website: bolor.me
- Email: [email protected]
License
MIT -- see LICENSE
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.