Memoriki

Memoriki

Personal knowledge base with real memory. Combines LLM Wiki (Andrej Karpathy) + MemPalace (MCP server).

Wiki gives structure. MemPalace gives memory.

The Problem

• LLM Wiki without MemPalace = library without a catalog. Search is just grep.
• MemPalace without Wiki = search engine without books. Semantic search over raw chunks.
• Together = structured knowledge + semantic search + entity graph.

Three Layers of Knowledge

| Layer | What it does | Tool | |-------|-------------|------| | Wiki | Structure, [[wiki-links]], YAML frontmatter, index | Markdown + Obsidian | | MemPalace Drawers | Semantic search over all content | mempalace_search | | MemPalace KG | Entity relationship graph with timestamps | mempalace_kg_query |

Architecture

memoriki/
  raw/                    # Your sources (articles, notes, transcripts)
  wiki/                   # LLM-generated wiki (LLM owns this entirely)
    index.md              # Page catalog - updated on every ingest
    log.md                # Operation log (append-only)
    entities/             # People, companies, products
    concepts/             # Ideas, patterns, frameworks
    sources/              # Summary page per source
    synthesis/            # Cross-cutting analysis, comparisons
  mempalace.yaml          # MemPalace config
  CLAUDE.md               # Schema and rules for the LLM
  idea-file.md            # Karpathy's original idea (reference)

Quick Start

# 1. Clone
git clone https://github.com/AyanbekDos/memoriki.git my-knowledge-base
cd my-knowledge-base

# 2. Install MemPalace
pip install mempalace
mempalace init .

# 3. Connect MemPalace to Claude Code
claude mcp add mempalace -- python -m mempalace.mcp_server

# 4. Drop your first source
cp ~/some-article.md raw/

# 5. Launch Claude Code and start ingesting
claude
# > Read raw/some-article.md and ingest it into the wiki

Operations

• Ingest - drop a file into raw/, tell the LLM to read and integrate it into the wiki
• Query - ask a question, LLM finds relevant pages and synthesizes an answer
• Lint - health check: contradictions, orphans, knowledge gaps

Works With

Any MCP-compatible LLM agent:

• Claude Code - use CLAUDE.md as-is
• OpenAI Codex - rename CLAUDE.md to AGENTS.md
• Cursor, Gemini CLI and other MCP-compatible tools

Use Cases

• Founders: customer discovery, interviews, competitors, pivots - all in one place
• Researchers: papers, articles, notes - wiki with compounding synthesis
• Students: lecture notes, books, projects - structured "second brain"
• Teams: Slack threads, meetings, decisions - AI-maintained wiki

License

MIT

Credits

• Andrej Karpathy - original LLM Wiki idea
• MemPalace - MCP server for semantic search and knowledge graph
• Claude Code - LLM agent

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: AyanbekDos
• Source: AyanbekDos/memoriki
• License: MIT

Install and usage instructions live in the source repository linked above.