AI Books MCP Server
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@AI Books MCP Server extend context from ./src/*.ts and explain authentication flow"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
AI Books MCP Server
Universal LLM Context Extension via Gravitational Memory Compression
Extend any LLM's context window by 15-60× while maintaining 100% data integrity. Built on quantum-inspired gravitational memory compression.
🚀 Features
Massive Context Extension: Extend LLM context 15-60× beyond native limits
100% Data Integrity: Cryptographic hash verification ensures perfect accuracy
Universal Compatibility: Works with Claude, GPT-4, Llama, and any LLM
Zero Configuration: Works out of the box with Claude Code
Lightning Fast: Query libraries in milliseconds
Memory Efficient: Compression ratios up to 1240× on dense technical content
Related MCP server: RLM MCP Server
📦 Installation
For Claude Code Users
npm install -g ai-books-mcp-serverThen add to your Claude Code MCP settings:
{
"mcpServers": {
"ai-books": {
"command": "ai-books-mcp-server"
}
}
}For Developers
git clone https://github.com/TryBoy869/ai-books-mcp-server.git
cd ai-books-mcp-server
npm install
npm run build🎯 Use Cases
1. Large Codebases
Create library from 100+ files → Query specific functionality → Get precise answers2. Research Papers
Compress 50 papers → Ask synthesis questions → Get citations + insights3. Documentation
Load entire docs → Natural language queries → Contextual answers4. Books & Long-form Content
Compress novels/textbooks → Ask thematic questions → Deep analysis🛠️ Available Tools
Core Tools
create_knowledge_library
Creates a compressed knowledge library from text.
{
name: "react-docs",
text: "...full React documentation...",
n_max: 15 // Optional: compression level (5-20)
}query_knowledge_library
Queries a library and retrieves relevant context.
{
library_name: "react-docs",
query: "How do hooks work?",
top_k: 8 // Optional: number of chunks (1-20)
}extend_context_from_files
Loads files and retrieves relevant context in one step.
{
file_paths: ["./src/*.ts"],
query: "Explain the authentication flow",
top_k: 8
}Management Tools
list_knowledge_libraries: List all librariesget_library_stats: Detailed statisticsdelete_knowledge_library: Remove a libraryverify_library_integrity: Check 100% integritysearch_documents: Search with relevance scores
📖 Example Usage
In Claude Code
User: Can you help me understand this React codebase?
Claude: [Calls create_knowledge_library with all React files]
[Creates library "react-project" with 245 chunks, 45× compression]
User: How does the authentication system work?
Claude: [Calls query_knowledge_library]
[Retrieves 8 most relevant chunks from authentication code]
[Provides detailed explanation with exact code references]Result
Instead of:
❌ "I can only see a few files at once"
❌ "The codebase is too large for my context"
You get:
✅ Full understanding of 100+ file codebases
✅ Accurate answers with specific code references
✅ Synthesis across multiple files
🧬 How It Works
Gravitational Memory Compression
Based on quantum physics' atomic orbital structure:
Text Chunking: Split documents into 200-300 word chunks
Hash Generation: SHA-256 hash for each chunk
Orbital Encoding: Map hash to gravitational states (quantum-inspired)
Compression: Achieve 15-60× reduction while maintaining retrievability
Verification: 100% integrity guaranteed via hash comparison
Technical Details
Algorithm: Gravitational bit encoding with n_max orbitals
Compression: 1240 discrete states per bit (n_max=15)
Retrieval: O(N) semantic similarity + O(1) hash lookup
Integrity: Cryptographic verification (SHA-256)
📊 Performance
Metric | Value |
Compression Ratio | 15-60× (typical) |
Data Integrity | 100% guaranteed |
Query Speed | < 100ms (1000 chunks) |
Max Library Size | Limited by RAM |
Chunk Retrieval | O(N) similarity scan |
🎓 Created By
Daouda Abdoul Anzize
Self-taught Systems Architect
40+ Open Source Projects
Specialization: Meta-architectures & Protocol Design
Portfolio: tryboy869.github.io/daa
GitHub: @TryBoy869
Email: anzizdaouda0@gmail.com
📄 License
MIT License - See LICENSE file
🤝 Contributing
Contributions welcome! Please:
Fork the repository
Create a feature branch (
git checkout -b feature/amazing)Commit your changes (
git commit -m 'Add amazing feature')Push to the branch (
git push origin feature/amazing)Open a Pull Request
🐛 Issues
Found a bug? Have a feature request?
🌟 Star History
If you find this useful, please star the repo! ⭐
🔗 Links
Built with ❤️ by Daouda Anzize | Extending LLM horizons, one library at a time
Available Tools
8 toolscreate_knowledge_libraryCreate Knowledge LibraryAIdempotent
Creates a new knowledge library by compressing text using gravitational memory. The text is split into chunks and compressed 15-60× while maintaining 100% data integrity. Perfect for large documents, codebases, or research papers.
Inputs: name (required), text (required), n_max. Returns: created_at, total_words, library_name, chunks_created, +1 more
delete_knowledge_libraryDelete Knowledge LibraryBDestructive
Permanently deletes a knowledge library.
Inputs: library_name (required). Returns: deleted, message, library_name
extend_context_from_filesExtend Context From FilesARead-onlyIdempotent
Loads multiple files, compresses them into temporary libraries, and retrieves relevant context for a given query. Perfect for quickly understanding large codebases or document sets without creating permanent libraries.
Inputs: query (required), file_paths (required), top_k. Returns: query, total_words, files_processed, extended_context, +2 more
get_library_statsGet Library StatisticsARead-onlyIdempotent
Retrieves detailed statistics for a specific knowledge library.
Inputs: library_name (required). Returns: n_max, created_at, updated_at, total_words, +5 more
list_knowledge_librariesList Knowledge LibrariesARead-onlyIdempotent
Lists all available knowledge libraries with their statistics.
Inputs: no named parameters. Returns: libraries, total_libraries
query_knowledge_libraryQuery Knowledge LibraryARead-onlyIdempotent
Queries a knowledge library and retrieves the most relevant chunks for a given query. Returns extended context that can be used to answer questions with much more detail than would fit in a normal LLM context window.
Inputs: query (required), library_name (required), top_k. Returns: query, context, total_words, library_name, +1 more
search_documentsSearch DocumentsARead-onlyIdempotent
Searches for relevant chunks in a knowledge library and returns previews with relevance scores.
Inputs: query (required), library_name (required), max_results. Returns: query, results, library_name, total_results
verify_library_integrityVerify Library IntegrityARead-onlyIdempotent
Verifies that all chunks in a library maintain 100% data integrity by checking hashes.
Inputs: library_name (required). Returns: all_verified, library_name, total_chunks, failed_chunks, +2 more
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 8 tools
Most tools have clearly distinct roles (create, list, delete, stats, verify), but query_knowledge_library and search_documents overlap in retrieving relevant chunks, requiring careful description reading to pick the right one.
The naming mostly follows a verb_noun pattern with consistent snake_case, but extend_context_from_files and search_documents deviate slightly from the knowledge_library-centric naming convention.
8 tools is well-scoped for a knowledge library server, covering creation, retrieval, management, integrity verification, and file-based context extration without unnecessary bloat.
The tool surface covers the core library lifecycle—create, list, query, delete—plus validation and file-based context loading. An update/add-to-existing-library operation is not present, but the domain appears adequately covered for typical use cases.
Maintenance
Related MCP Connectors
Memory that reasons: continual learning for stateful agents. Better context, fewer tokens.
Persistent memory and knowledge graphs for AI agents. Hybrid search, context checkpoints, and more.
Deterministic memory & code intelligence MCP for 15 IDEs (100% LongMemEval in 12.9ms, 0 LLM tokens).
Compress coding-agent context ~64%. Hosted Neural Keep MCP for Cursor/Claude/Codex.
Related MCP Servers
- AlicenseAqualityCmaintenanceExtends large language model context length up to 60 times using gravitational memory, designed for Claude Code and Anthropic AI models.813 npm2MIT
- AlicenseAqualityDmaintenanceEnables any LLM to process arbitrarily long contexts through recursive decomposition, without requiring external LLM APIs.1718 npm15MIT
- AlicenseNot gradedqualityDmaintenanceVirtual Infinite Context for Agents and LLMs - maintains a continuous rolling context window, surfacing relevant memories while respecting token budgets.6MIT
- AlicenseNot gradedqualityDmaintenanceLocal-first, cross-session context store that reduces token usage by saving facts, decisions, and preferences, and recalling them in later sessions with token-efficient ranking and compression.6 npmMIT