Skip to main content
Glama

ProjectMind MCP

Give your AI coding assistant a brain. Persistent memory, semantic code search, a symbol-level call graph, and project intelligence โ€” all running locally with no API keys required.

CI Python License Version PyPI

ProjectMind is an open-source MCP (Model Context Protocol) server that supercharges AI assistants like Claude, Zencoder, and Cursor with long-term project memory and intelligent codebase search.

๐Ÿค– This project was built with AI โ€” designed, coded, debugged, and documented using AI-assisted development from day one.


Why ProjectMind?

Every time you start a new AI session, your assistant forgets everything about your project. ProjectMind solves this:

  • No more re-explaining your architecture every session โ€” memory persists in .ai/memory.md

  • Semantic code search that understands what code does, not just what it's named

  • Symbol graph โ€” ask "who calls this function?" and get exact file:line answers

  • Dependency graph analysis to understand how modules connect

  • Works 100% locally โ€” your code never leaves your machine

New in v0.9.0: โœ๏ธ AI-authored annotations (semantic search without embeddings), ๐Ÿ“ฆ lightweight core with the vector stack as an optional [vector] extra, ๐Ÿš€ one-line install from PyPI.

๐Ÿ“ See CHANGELOG.md for the full release history.


Related MCP server: nodespace-mcp

Features

๐Ÿง  Persistent Project Memory

Save architectural decisions, tech stack notes, and context that survives across sessions. The AI reads this at the start of every conversation.

Hierarchical memory access avoids dumping everything at once:

  • read_memory_index() โ€” section headings only (cheap)

  • read_memory_section(name) โ€” expand only the section you need

  • search_memory(query) โ€” relevance-ranked retrieval of the memory blocks most relevant to a task (keyword-scored), not just the head of the file

All memory I/O is UTF-8 โ€” notes in any language survive round-trips on any OS.

Queries escalate through tiers only when the previous one is insufficient โ€” large repositories never trigger a cold load just to answer a path lookup.

Tier

Engine

When used

Typical latency

L0 Manifest

.ai/manifest.json (paths + whole-file symbols)

always

< 50 ms

L0 Annotations

.ai/annotations.json (AI-written summaries)

always

< 5 ms

L1 Symbol

.ai/symbol_graph.json (AST symbol names)

when a prebuilt graph exists

< 10 ms

L1 BM25

rank-bm25 lexical index (chunks + annotations)

when L0 is weak

~ 100 ms

L2 Vector

ChromaDB + sentence-transformers (optional [vector] extra)

only on intent='semantic'/'deep'

first call ~ 30 s, then cached

Use the unified query(text, intent, n_results) tool with one of:

  • overview โ€” L0 only (paths + top-level symbols)

  • lookup โ€” L0 + symbols + BM25 (keyword/lexical)

  • semantic โ€” escalates to embeddings if the lexical signal is weak

  • deep โ€” all tiers with relaxed thresholds

Hits from every tier are fused with scale-invariant Reciprocal Rank Fusion (normalized to 0..1), so no single tier can dominate the ranking by raw score magnitude โ€” results corroborated across tiers float to the top.

โœ๏ธ AI-Authored Annotations โ€” semantic search without embeddings

The killer pattern: your AI assistant already understands your code โ€” so let it write the search index. As the assistant works on files, it saves 1-2 sentence summaries + keywords via save_annotation(). Annotations are indexed into BM25 and a dedicated query tier, so natural-language queries land on the right files with plain keyword search:

save_annotation("src/auth/session.py",
    "Session lifecycle: creates, refreshes and revokes OAuth sessions",
    keywords="login, oauth, token refresh")

query("where are sessions revoked?")  โ†’  src/auth/session.py  (L0_annot tier)
  • list_unannotated_files() โ€” coverage report: files missing or with stale annotations (file changed since)

  • get_annotations(path) โ€” review what the index "knows" about a file

  • Stored in human-readable .ai/annotations.json, UTF-8, atomic writes

  • This is why the vector stack is now optional: a smart model + cheap precise tools beats a small embedding model guessing

๐Ÿงฌ Symbol Graph (AST-level call/inherit/implement graph)

tree-sitter parses every source file into a graph of symbols, not files. Symbols are keyed by qualified id (path::Class.method), so same-named symbols across files stay distinct.

  • find_symbol("UserService") โ€” where is it defined? (no embedding model needed, answers in milliseconds)

  • get_symbol_relations("save", relation="callers") โ€” who calls it, with file:line

  • Relations: callers, callees, implementors, subclasses, bases, usages, info

  • Inheritance extraction works across Python, JS, TS, Java, Ruby; builtin-call noise (print, len, pushโ€ฆ) is filtered out

  • The graph is rebuilt automatically by the background indexer and persisted to .ai/symbol_graph.json

Search your codebase by meaning, not just text. Powered by a local sentence-transformers model โ€” no OpenAI key needed.

"find authentication middleware"  โ†’  finds auth code even if it's named differently

๐ŸŒณ AST-Aware Chunking

Unlike naive text splitters that cut code in the middle of a function, ProjectMind uses tree-sitter to parse source files into exact syntax units:

  • Functions and methods are indexed as individual, self-contained chunks

  • Class methods get a # Class: ClassName context prefix for better search relevance

  • Rich metadata per chunk: symbol_type, symbol_name, class_name, line_start, line_end

  • Supports: Python, JavaScript, TypeScript, TSX, Java, Go, Rust, Ruby

  • Graceful fallback to text splitting for unsupported file types

๐Ÿ•ธ Dependency Graph Intelligence

  • Traverse import relationships up to 5 levels deep

  • Find related files via shared dependency clustering

  • Discover the shortest path between any two modules

  • Identify entry points and orphaned modules

  • Monorepo-aware JS/TS resolution โ€” follows tsconfig/jsconfig path aliases (@/...) and workspace/package imports; multi-line, dynamic import() and side-effect imports are all detected

  • Python src/-layout & relative-import resolution โ€” absolute imports resolve through src//lib/ roots, dotted (utils.helpers) and relative (./..) imports resolve to the right files

  • Cached import graph (120 s TTL) with a precomputed reverse graph โ€” impact/cluster analysis is O(E), not O(NยฒยทE)

โšก Instant Project Exploration (no indexing needed)

  • get_project_overview() โ€” manifest-first; tech stack, git info, file stats in < 1 second

  • explore_directory(path) โ€” browse project tree level by level

  • get_file_summary(path) โ€” imports, classes, functions, git history

โšก Hybrid Search (BM25 + Vector)

Two search engines combined via Reciprocal Rank Fusion (RRF):

  • BM25 catches exact keyword matches โ€” finds getUserById when you type exactly that

  • Vector search catches semantic matches โ€” finds auth code even if named differently

  • RRF merges both ranked lists for best-of-both-worlds results

  • Automatic fallback to pure vector search when the BM25 index is not ready

  • The BM25 index is persisted as JSON (.ai/bm25_index.json) โ€” never pickle, so indexing a third-party repo can't execute untrusted input

๐Ÿ”„ True Incremental Indexing

index_changed_files re-indexes only what changed โ€” and cleans up after itself:

  • A changed file's old chunks are deleted before the new ones land (renamed/removed symbols don't haunt search results)

  • Chunks of deleted files are removed from ChromaDB, BM25 and the metadata

  • BM25 is patched in memory per file and rebuilt once โ€” no full-database fetch per small change

  • Failed writes are never silent: metadata isn't saved, so affected files retry on the next run

๐Ÿƒ Background Indexing

index_codebase() returns instantly and indexes in a daemon thread; session_init auto-starts it when the index is missing. Poll get_index_progress() for a live progress bar with ETA. Switching projects mid-run cancels the old job safely โ€” no cross-project contamination.

๐Ÿฉบ Self-Healing Maintenance Daemon

A background thread keeps the index lean without user intervention. State persists in .ai/maintenance_state.json.

Task

Trigger

Action

manifest_refresh

every 5 min

rebuild L0 manifest if files changed

stale_gc

hourly

delete embeddings for files removed from disk

db_compaction

daily

VACUUM ChromaDB SQLite when > 200 MB

log_truncate

every 6 h

truncate projectmind.log when > 8 MB

model_unload

every 5 min

release sentence-transformers after 1 h idle (env-tunable)

cache_pressure

every minute

drop file/query caches when RSS > 500 MB

Inspect with maintenance_status(); force a sync run with maintenance_run(); aggressively clean the index with prune_index(force=True).

๐Ÿ“Š Code Quality Metrics

Cyclomatic complexity, pylint scores, test coverage tracking โ€” all queryable via MCP tools. Both analyze_code_complexity and analyze_code_quality accept mode='quick' (default, fast) or mode='deep' (wider scan). pylint runs in a subprocess with a wall-clock budget, so these tools return partial results instead of timing out.

โšก Lazy session_init (no 30 s timeouts)

session_init never loads the embedding model; it returns the project root + manifest + memory index in well under a second even on multi-GB repositories. The vector store is loaded only when an intent='semantic' or 'deep' query actually needs it.


Quick Start

# Claude Code
claude mcp add --scope user Memory -- uvx projectmind-mcp

The default install is lightweight (a few MB โ€” BM25 keyword search, AI annotations, symbol graph, memory). To add the optional embedding/vector tier (~500 MB, fully local):

claude mcp add --scope user Memory -- uvx --from "projectmind-mcp[vector]" projectmind-mcp

Other MCP clients (Claude Desktop / Zencoder / Cursor) โ€” mcp.json:

{
  "mcpServers": {
    "Memory": {
      "command": "uvx",
      "args": ["projectmind-mcp"]
    }
  }
}

From source (development)

git clone https://github.com/Nik0lay1/project-mind-mcp.git
cd project-mind-mcp
python -m venv .venv

# Windows                          # macOS/Linux
.venv\Scripts\pip install -e ".[vector,dev]"   # .venv/bin/pip install -e ".[vector,dev]"

Then point your MCP client at .venv/Scripts/python.exe mcp_server.py (or use the projectmind-mcp console script from the venv).

3. Bootstrap a session

Ask the AI (once per session):

Memory__session_init(project_path="<absolute path to your project>")

If the index doesn't exist yet, background indexing starts automatically โ€” check Memory__get_index_progress(). For manual full re-indexing of large projects:

# Windows
.venv\Scripts\python.exe run_index.py

# macOS/Linux
.venv/bin/python run_index.py

Available Tools (45+)

Category

Tools

Session

session_init, health, set_project_root

Memory

read_memory, read_memory_index, read_memory_section, search_memory, update_memory, clear_memory, save_memory_version

Search

query (tier-aware), search_codebase, search_for_feature, search_architecture, search_for_errors, search_with_dependencies

Context brief

get_context_brief โ€” ranked, budget-packed brief for a coding task: hybrid retrieval + 1-hop import graph + file skeletons + git recency; zero LLM calls (the "scout pyramid" layer 0)

Annotations

save_annotation, get_annotations, list_unannotated_files

Symbols

find_symbol, get_symbol_relations (callers / callees / implementors / subclasses / bases / usages)

Exploration

get_project_overview, explore_directory, get_file_summary

Dependencies

get_file_relations, get_dependencies_with_depth, get_module_cluster, find_dependency_path, analyze_change_impact

Indexing

index_codebase (background by default), index_changed_files, get_index_progress, get_index_stats, prune_index

Git

ingest_git_history, get_recent_changes_summary, auto_update_memory_from_commits

Quality

analyze_code_complexity, analyze_code_quality, get_test_coverage_info

Maintenance

maintenance_status, maintenance_run

Project

detect_project_conventions, generate_project_summary

Full reference: docs/api/tools-reference.md


How It Works

Your Project
     โ”‚
     โ–ผ
ProjectMind MCP Server
     โ”‚
     โ”œโ”€โ”€ .ai/memory.md                โ† persistent notes & decisions (UTF-8)
     โ”œโ”€โ”€ .ai/annotations.json         โ† AI-written file summaries (search tier)
     โ”œโ”€โ”€ .ai/manifest.json            โ† L0: paths, symbols, modules (โ‰ค200 KB)
     โ”œโ”€โ”€ .ai/symbol_graph.json        โ† L1: AST call/inherit/implement graph
     โ”œโ”€โ”€ .ai/bm25_index.json          โ† L1: lexical index (JSON, not pickle)
     โ”œโ”€โ”€ .ai/vector_store/            โ† L2: ChromaDB embeddings (local)
     โ”œโ”€โ”€ .ai/index_metadata.json      โ† tracks changed files (mtime)
     โ”œโ”€โ”€ .ai/index_progress.json      โ† live background-indexing progress
     โ”œโ”€โ”€ .ai/maintenance_state.json   โ† self-healing daemon schedule
     โ””โ”€โ”€ .ai/.indexignore             โ† per-project ignore patterns
     โ”‚
     โ–ผ
AI Assistant (Claude / Zencoder / Cursor)

Embedding model: flax-sentence-embeddings/st-codesearch-distilroberta-base

  • Trained specifically on code (CodeSearchNet dataset)

  • ~130 MB, runs fully locally on CPU

  • No API keys, no data sent anywhere

Search pipeline: manifest + symbol graph + BM25 (keyword) + ChromaDB (semantic) โ†’ Reciprocal Rank Fusion โ†’ top-N results


Requirements

  • Python 3.10 โ€“ 3.12

  • ~500 MB disk (model + dependencies)

  • Works on Windows, macOS, Linux


Configuration

All settings in config.py:

Setting

Default

Description

MODEL_NAME

flax-sentence-embeddings/st-codesearch-distilroberta-base

Embedding model

CHUNK_SIZE

1500

Characters per chunk

MAX_FILE_SIZE_MB

10

Skip files larger than this

MAX_MEMORY_MB

100

Memory limit for indexing batch

IMPORT_GRAPH_MAX_FILES

8000

Max files scanned when building the import graph

TOOL_SOFT_BUDGET_SECONDS

20

Wall-clock budget for analysis tools โ€” they return partial results instead of timing out

Override via environment variables:

PROJECTMIND_MAX_FILE_SIZE_MB=5
PROJECTMIND_MAX_MEMORY_MB=200
PROJECTMIND_IMPORT_GRAPH_MAX_FILES=20000
PROJECTMIND_SYMBOL_GRAPH_MAX_FILES=8000
PROJECTMIND_TOOL_BUDGET_SECONDS=45
PROJECTMIND_MODEL_IDLE_UNLOAD_SECONDS=3600

Custom ignore patterns: create .indexignore in the project root (fallback: .ai/.indexignore), same substring syntax as shown in the generated default.


Project Structure

mcp_server.py           โ† all MCP tool definitions
config.py               โ† configuration + path validation
manifest.py             โ† L0 lightweight project manifest
symbol_graph.py         โ† AST symbol graph (call/inherit/implement), format v3
query_router.py         โ† tier-aware query() router (L0 โ†’ L1_symbol โ†’ L1 โ†’ L2)
maintenance.py          โ† self-healing background daemon
background_indexer.py   โ† non-blocking indexing with live progress
vector_store_manager.py โ† ChromaDB wrapper + hybrid search (L2)
bm25_index.py           โ† BM25 keyword index + RRF fusion (L1, JSON persistence)
codebase_indexer.py     โ† file scanning & AST-aware chunking
ast_splitter.py         โ† tree-sitter parser (9 languages, thread-safe)
code_intelligence.py    โ† import graph, complexity analysis, cached graphs
memory_manager.py       โ† persistent memory read/write (UTF-8, atomic)
incremental_indexing.py โ† change tracking (mtime) + atomic writes
context.py              โ† dependency injection (thread-safe lazy init)
run_index.py            โ† helper script for manual re-indexing

Contributing

Issues and PRs are welcome. This is an open project โ€” built in the open, improved in the open.

pip install -e ".[dev]"
pytest tests/
ruff check .

CI runs ruff, black, mypy and the unit suite on Python 3.10โ€“3.12 (see .github/workflows/ci.yml).


License

MIT


Built with AI assistance โ€” used throughout development for coding, debugging, refactoring, and documentation.

A
license - permissive license
-
quality - not tested
A
maintenance

Maintenance

โ€“Maintainers
โ€“Response time
1dRelease cycle
3Releases (12mo)
Commit activity

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/Nik0lay1/project-mind-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server