| name | anime-mori |
| description | Persistent codebase intelligence. Learns patterns, conventions, and architecture from your codebase via tree-sitter AST parsing (12 languages) and remembers across sessions. MCP server with 13 tools for semantic search, pattern prediction, file routing, and project blueprints. |
| license | MIT |
| metadata | {"version":"0.6.0","runtime":"node>=18","dependencies":{"required":["@modelcontextprotocol/sdk","@surrealdb/node","tree-sitter (via rust-core NAPI bindings)"],"optional":["@xenova/transformers"]},"env_vars":{"required":[],"optional":["ANIME_MORI_DB_PATH"]}} |
Anime Mori: Persistent Codebase Intelligence
"Your AI pair programmer finally has a notepad that doesn't get wiped clean."
Overview
Anime Mori is the codebase memory skill of the Cognitive Construct. It solves session amnesia — the problem where AI assistants forget your architecture, naming conventions, and design decisions every time you start a new session.
Built on Rust (tree-sitter AST parsing) + TypeScript (MCP server + orchestration), it learns once and remembers forever via SurrealDB persistence.
Architecture
AI Tool (Claude/Copilot)
│ MCP Protocol (stdio)
▼
TypeScript Server — 13 tools, 4 categories
│ NAPI bindings
▼
Rust Core — tree-sitter, pattern learning, semantic engine
│
▼
Storage — SurrealDB (persistent vectors) + SQLite (structured data)
Commands
learn <path>
Analyze and learn from a codebase. Extracts patterns, architecture, and conventions.
npx anime-mori learn ./my-project
server
Start the MCP server for AI tool integration.
npx anime-mori server
search <query>
Multi-mode codebase search (semantic, text, or pattern-based).
npx anime-mori search "authentication middleware"
analyze <path>
Analyze files/directories with concepts, patterns, and complexity metrics.
insights <query>
Query learned semantic concepts and relationships.
patterns
Get pattern recommendations with examples from your codebase.
predict
Predict implementation approach based on your coding history.
status
System health check.
MCP Tools (13)
| Category | Tools | Description |
|---|
| Core Analysis (2) | analyze_codebase, search_codebase | File/directory analysis, multi-mode search |
| Intelligence (7) | learn_codebase_intelligence, get_project_blueprint, get_semantic_insights, get_pattern_recommendations, predict_coding_approach, get_developer_profile, contribute_insights | Deep learning, blueprints, predictions |
| Automation (1) | auto_learn_if_needed | Smart auto-learning with staleness detection |
| Monitoring (3) | get_system_status, get_intelligence_metrics, get_performance_status | Health, analytics, diagnostics |
Language Support
Native AST parsing via tree-sitter (12 languages):
TypeScript, JavaScript (JSX/TSX), Python, Rust, Go, Java, C, C++, C#, Svelte, SQL, PHP
Synergies
- → Encyclopedia: Vendored at
encyclopedia/scripts/anime-mori/ — encyclopedia can invoke Anime Mori's MCP tools for codebase context during knowledge retrieval
- → Inland Empire: Learned patterns and conventions feed into the subconscious memory layer
Files
src/ — TypeScript source (CLI, MCP server, engines, storage)rust-core/ — Rust NAPI bindings (tree-sitter parsers, analyzers)src/mcp-server/server.ts — MCP server entry pointsrc/cli/cli.ts — CLI entry pointsrc/engines/ — Semantic and pattern analysis enginessrc/storage/ — SurrealDB persistence layerREADME.md — Full documentation with setup guides
