MSCodebase Intelligence

AI-powered semantic code search for Zed IDE — deep code analysis MCP server

Python 3.14+ License: MIT MCP Zed CI Tests

Features • Quick Start • Tools • Documentation • Installation • Architecture • Contributing • Security

Last updated: 2026年08月16日

🎯 Positioning

MSCodeBase Intelligence is an MCP server for Zed IDE that gives AI assistants deep understanding of the entire codebase: semantic search, call graph, project memory, diagnostics.

This is not an LSP server or a replacement for the editor's built-in autocomplete. It's a "code intelligence" layer on top of the editor:

┌─────────────────────────────────────────────────────┐
│ Zed IDE │
│ ┌───────────────────────────────────────────────┐ │
│ │ LSP (built-in autocomplete, │ │
│ │ inline hints, diagnostics) │ │
│ └───────────────────────────────────────────────┘ │
│ │ │
│ ▼ │
│ ┌───────────────────────────────────────────────┐ │
│ │ MSCodeBase (MCP server) │ │
│ │ · Semantic search across the codebase │ │
│ │ · Call graph & impact analysis │ │
│ │ · Project memory (ADR, tech debt) │ │
│ │ · Self-diagnostics and self-healing │ │
│ │ · 65 tools for AI assistant │ │
│ └───────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────┘

What you get

Feature	MSCodeBase	Standard LSP (pyright/pylsp)
🔍 Semantic search (BM25 + Vector + Reranker)	✅	❌
🧠 Call graph + impact analysis	✅	❌
🗃️ Project memory (ADR, known issues)	✅	❌
🏥 Self-diagnosis + self-healing	✅	❌
🔎 Cross-repo search	✅	❌
🤖 RAG answer generation (mode=ask)	✅	❌
🔬 Search explainability (per-stage score trace)	✅	❌
🏛️ Architecture drift detection (chain/circular/hub)	✅	❌
✅ Claim verification (agent fact-checking vs code)	✅	❌
✏️ Inline autocomplete	❌	✅
🏷️ Inlay hints	❌	✅

LSP: Hybrid Rename Only

MSCodeBase uses LSP only for codebase(action="rename") — the LSP client (src/core/lsp_client.py) spawns pyright-langserver for precise cross-file rename, with graceful fallback to SymbolIndex (Tree-sitter) on timeout. All other functionality is implemented through 61 MCP tools.

The standalone LSP server (src/lsp_main.py) was experimental and does not work in Zed — see LSP_WONTFIX.md.

Platforms

Designed and tested on Windows. macOS and Linux should work but have not been validated officially.

Languages

Language	Parsing	Call Graph	Data Flow (ASSIGNED_FROM)
Python	✅	✅	✅
TypeScript	✅	✅	✅
TSX	✅	✅	✅
Rust	✅	✅	✅
Go	✅	✅	✅
JavaScript	✅	✅	✅
Java	✅	✅	✅
C#	✅	✅	✅
Ruby	✅	✅	✅
PHP	✅	✅	✅
Kotlin	✅	✅	✅
Swift	✅	✅	✅
C	✅	✅	✅
C++	✅	✅	✅
Scala	✅	✅	✅
Dart	✅	✅	✅
Shell / Bash	✅	✅	❌ (grammar without RHS-field)
SQL	✅ (context)	❌	❌
YAML	✅ (context)	❌	❌
TOML	✅ (context)	❌	❌
HTML	✅ (context)	❌	❌
CSS	✅ (context)	❌	❌
HCL / Terraform	✅ (context)	❌	❌

✨ Features

Feature	Description
🔍 Unified Search	`search_code(query, mode, intent_hint)` — single tool: fast/quality/deep/context/ask/auto
🧠 Intelligence Layer	16 high-level `intel_*` tools: self-diagnostics, topology, memory, error prediction
🌐 Cross-repo Search	Search across multiple projects with `@mention` syntax
🌳 Call Graph	Full call graph: definition + callers + callees + impact analysis
🏗 Structural Search	13 AST patterns (class_inheritance, async_function, decorator, etc.)
🔎 Context Search	Find similar code — paste a fragment, get semantic duplicates
🪣 Multi-Bucket RAG	Code/docs buckets, soft weighting, intent_hint (code/docs/auto)
🤖 mode=ask	RAG answer generation via phi-4 (server profile)
💾 LanceDB v2	Vector DB with per-project isolation (incremental BM25 reindex)
🛡 Rate Limiting	DebounceBatch + CircuitBreaker — protection against VFS loops
🏥 Self-Diagnosis	`get_health_report` + `index_health` — full check and recovery
🧪 Clean Architecture	DI Container (18 services), 65 tools (31 core + 16 intel + 14 inline + 4 dev), ~1677 tests
🪟 Multi-Window	`ProjectIndexerRegistry` — isolated Indexer per project, LRU 5, ResourceMonitor throttle
✏️ Write Tools	`codebase(action=...)` — unified hub: rename, move, delete, replace, insert, ack
⚡ Meta-Patching	LanceDB `move_chunks_metadata` — file_path rename without re-embedding (50ms vs 5s)
🔗 Data Flow Graph	`ASSIGNED_FROM` edges track variable assignments. Unified Walker + Conditional Flow (if/for/while/try). 29 edge types in PropertyGraph.
⚙️ SYSTEM_PROFILE	`light` (sync) / `server` (async with phi-4)
🎯 MMR Diversification	Maximal Marginal Relevance (λ=0.6) after RRF — removes duplicates while preserving relevance. 0.3ms for 50 docs.
🧠 Auto Intent Detection	Keyword-based auto-detection of code/docs intent from query text. No manual `intent_hint` required.
📖 Extended Synonyms	39 synonym groups (auth↔login, function↔method, cache↔buffer, etc.) — bridges the gap between user terminology and code.

🚀 Quick Start

Install the mscodebase-intelligence extension in Zed, then:

cd D:\Project\MSCodeBase
python install.py
# Quick sync (code only, no prompts):
python install.py --sync
# CI mode (no prompts, fail fast):
python install.py --yes
# Skip model downloads:
python install.py --skip-models
# Restart Zed (File → Quit → reopen)
# Verify: intel_get_runtime_status()

install.py does:

Copies 39+ source files to the extension directory
Installs Python dependencies
Downloads llama-server.exe + GGUF reranker model (bge-reranker-v2-m3). The embedder (multilingual-e5-small INT8) is an ONNX model downloaded separately.
Configures MCP in Zed's settings.json

Providers

MCP auto-selects the best available provider (in priority order):

llama.cpp GGUF (native, preferred) → ONNX INT8 (in-process fallback) → LM Studio (if running) → BM25 only
 ~1.7 GB RAM (llama-server) ~0.5 GB RAM ~6 GB RAM no embeddings
 e5-small GGUF (384dim) e5-small INT8 (384dim) external API

Embedding runs via llama.cpp (llama-server.exe, preferred; ONNX in-process preload is canceled when llama.cpp is available). The reranker runs as a separate llama-server.exe process serving the BGE-M3 GGUF model. ONNX INT8 / LM Studio are fallback providers if llama.cpp is unavailable.

Benchmarks: docs/research/2026-07-10-final-benchmark.md

📚 Documentation Map

Document	Description	Audience	Languages
docs/en/INSTALL.md	Installation, setup, uninstall	Users	🇬🇧 🇷🇺 🇨🇳
docs/en/ARCHITECTURE.md	Clean Architecture, Layers, DI	Developers	🇬🇧 🇷🇺 🇨🇳
docs/en/ARCHITECTURE_DEEP.md	Deep architecture: pipeline, lifecycle, comparison	Architects	🇬🇧 🇷🇺 🇨🇳
docs/en/SEARCH_PIPELINE.md	Search pipeline: BM25 → RRF → Reranker	Developers	🇬🇧 🇷🇺 🇨🇳
docs/en/GRACEFUL_DEGRADATION.md	5 levels of graceful degradation (llama.cpp → ONNX → BM25)	DevOps	🇬🇧 🇷🇺 🇨🇳
docs/en/ARCHITECTURE_LAYERS.md	10 runtime layers	Architects	🇬🇧 🇷🇺 🇨🇳
docs/en/FAQ.md	Frequently Asked Questions	All	🇬🇧 🇷🇺 🇨🇳
docs/en/TELEMETRY.md	Metrics, ETA, data collection	DevOps	🇬🇧 🇷🇺 🇨🇳
docs/en/investigations/ONNX_SESSION_REPORT.md	Full ONNX migration, 7 fixes, benchmarks	Support	🇬🇧
docs/en/investigations/LSP_WONTFIX.md	LSP on Windows investigation (WONTFIX)	Support	🇬🇧 🇷🇺 🇨🇳
docs/en/ZED_WINDOWS_QUIRKS.md	Windows specifics, Restricted Mode	Windows users	🇬🇧 🇷🇺 🇨🇳
docs/en/CHANGELOG.md	Version history	All	🇬🇧 🇷🇺 🇨🇳
docs/en/CONTRIBUTING.md	How to contribute, PRs	Contributors	🇬🇧 🇷🇺 🇨🇳
docs/en/SECURITY.md	Security policy, vulnerabilities	Security	🇬🇧 🇷🇺 🇨🇳
AGENTS.md	AI Agent system rules	AI Agent	🇬🇧
SECURITY.md	Security policy, reporting vulnerabilities	Security	🇬🇧
CODE_OF_CONDUCT.md	Community standards	Contributors	🇬🇧
CONTRIBUTING.md	How to contribute (root-level)	Contributors	🇬🇧
KNOWN_ISSUES.md	Known issues & technical debt registry	All	🇬🇧

All documents are cross-referenced. Available in 3 languages: English, Русский, 中文.

Research & Writeups

Deep-dives into specific technical findings from building this project:

PageRank vs RAG on a Real Codebase: Corrected Numbers, and What I Almost Got Wrong Twice — comparing retrieval methods, with two rounds of honest self-correction
I Asked One AI to Fact-Check Another AI's Audit of My Own Code
The Silent Vector Contamination Bug: Why Your Concurrent Embeddings Might Be Lying to You

🔧 MCP Tools (65 total)

64 = 63 base + execute_script (регистрируется при MSCODEBASE_EXECUTE_SCRIPT_ENABLED=true). Без флага — 63 (30 core + 16 intel + 13 inline + 4 dev).

Core Search

Tool	When to Use
`search_code(query, mode, filter_layer, intent_hint)`	Main search tool. `mode="auto"` / `"fast"` / `"quality"` / `"deep"` / `"context"` / `"ask"`. `intent_hint="code"` / `"docs"` / `"auto"` — soft bucket weighting. `filter_layer="core"` — search within specific architecture layer
`structural_search(pattern)`	AST search: `class_inheritance`, `async_function`, `function_with_decorator` and more
`cross_repo_search(query @repo)`	Search across multiple projects (mono-repo)
`cross_project_deps(action)`	Cross-project dependency graph: `graph` / `deps` / `cycles` / `impact`
`get_symbol_info(query)`	Call Graph: callers, callees, impact files
`execute_script(code, timeout, args)`	Sandboxed Python execution (3-layer). AST validation + runtime `__import__` wrapper + subprocess isolation. Audit-logged. Returns `{stdout, stderr, exit_code, duration_ms, truncated, timed_out}`
`impact_analysis(symbol)`	Symbol change impact analysis (risk score, depth)

LSP Analysis (basedpyright)

Tool	When to Use
`lsp_find_references(file_path, line, col, symbol_name)`	Exact AST references (all usages of a symbol) via Zed's bundled basedpyright. `line`/`col` are 0-based (LSP); `col=-1` auto-detects via `symbol_name`
`lsp_find_definition(file_path, line, col, symbol_name)`	Jump to symbol definition (declaration) — precise, language-server-grade
`lsp_document_symbols(file_path)`	File structure tree: classes/functions/variables with positions

Index Management (via `codebase(action="index", ...)`)

Action	When to Use
`codebase(action="index", path="status")`	Index status: chunks, files, symbols (`get_index_status`)
`codebase(action="index", path="progress")`	Indexing progress (phase, percent)
`codebase(action="index", path="project_dir", project_root=...)`	Start full project indexing (`index_project_dir`; prefer async `intel_trigger_reindex`)
`codebase(action="index", path="timeline")`	Indexing history by date
`codebase(action="index", path="health")`	Index diagnostics and self-recovery (`index_health`)
`notify_change(file_path)`	Force index update for a file (via DebounceBatch) — inline tool
`generate_chunk_summaries(root)`	LLM-generated descriptions for code chunks
`scan_changes(project_root)`	Architectural diff — analyze changes since last baseline

System & Diagnostics

Tool	When to Use
`get_health_report()`	Full self-diagnosis: index, embedder, logs, synchronization
`get_logs(project_root)`	Latest errors and warnings from project logs
`read_live_file(path)`	Read file from LSP memory (including unsaved changes)

Analytics

Tool	When to Use
`get_hotspots(project_root)`	Hotspots — files with high bug rate
`get_repo_rank(project_root, top_k)`	Symbol importance ranking (PageRank on call graph)
`get_bug_correlation(project_root)`	Bug-change correlation analysis
`get_repo_map(project_root)`	Project map: file tree + key symbols
`graph_query(action="related", target=path)`	Files related via co-change / bug correlation (via `related` action)
`graph_query(action, target)`	Graph queries: `impact` / `feature` / `deps` / `tests` / `cypher` / `flow` / `drift` / `verify`
`find_similar_bugs(error)`	Find similar bugs from history by error text

Git & History (via `codebase(action="git", ...)`)

Action	When to Use
`codebase(action="git", path="log", ...)`	Semantic commit history (`get_commit_history`)
`codebase(action="git", path="history", ...)`	Change history for a specific file
`codebase(action="git", path="branch")`	Branch info + index status (`get_branch_info`)

Lifecycle & Verification

Tool	When to Use
`submit_background_task(type, root)`	Run long tasks: `bug_correlation` / `build_knowledge_graph` / `full_analysis`
`get_task_status(task_id)`	Background task status
`verify_action(action_type)`	Verification: `file_write` / `git_commit` / `git_push` / `index_sync`

Write Tools — `codebase(action=...)`

Action	When to Use
`codebase(action="rename", old, new, apply)`	Rename symbol across all files (preview/apply, collision check)
`codebase(action="move", symbol, to_file, apply)`	Move symbol to another file (preview/apply, import updates)
`codebase(action="safe_delete", symbol, force, apply)`	Safe delete with reference check (force mode)
`codebase(action="replace", symbol, new_code, apply)`	Replace function/class body (preview/apply)
`codebase(action="insert_before", anchor, new_code, apply)`	Insert code before anchor symbol (preview/apply)
`codebase(action="insert_after", anchor, new_code, apply)`	Insert code after anchor's body (preview/apply)
`codebase(action="ack_impact", file_path)`	Acknowledge impact for modification guard

Intelligence Layer (intel_*) — 16 High-Level Tools

Tool	What it does
`intel_get_runtime_status()`	Aggregated health status: embedder, index, resource usage
`intel_trigger_reindex()`	Fire-and-forget reindexing (does not block Zed)
`intel_get_job_status(job_id)`	Background task progress
`intel_code_topology(symbol)`	Call graph + module topology (< 2 sec)
`intel_get_project_memory()`	Project memory map: ADR, known_issues, tech_debt
`intel_log_incident(...)`	Log an incident to project history
`intel_analyze_incident(error)`	Find similar incidents + ready-made solutions
`intel_add_memory_node(section, data)`	Add a record to project memory
`intel_get_hotspots()`	Top-5 files with highest bug load
`intel_predict_root_cause(error)`	Predict root cause from logs + history
`intel_get_telemetry(days)`	Per-tool telemetry, resource usage, LLM stats
`intel_auto_collect_adrs(max_commits)`	Auto-generate ADRs from commit history
`intel_reset_index()`	Delete and rebuild index from scratch
`intel_retract_memory_node(node_id, reason)`	Retract a memory node (ACTIVE/VERIFIED → REFUTED, reason required)
`intel_restore_memory_node(node_id, reason)`	Restore a memory node from REFUTED (manual return, ADR-0002/0003)
`intel_supersede_memory_node(node_id, reason, new_node_id)`	Mark a node as SUPERSEDED — replaced by a newer fact

intel_tool_health(), intel_explain_project_state(), intel_get_project_context() — see Diagnostic Tools below.

Dev Tools (4)

Tool	What it does
`generate_docs(project_root)`	Generate Markdown docs from PropertyGraph (DEPRECATED — use auto_update_docs)
`bump_version(project_root, part, dry_run)`	Bump project version + update CHANGELOG
`auto_update_docs(project_root, action)`	Auto-update documentation: update/check
`install_git_hooks(project_root, action)`	Install pre-commit hooks: install/uninstall/status

Diagnostic Tools (7)

Tool	What it does
`debug_runtime_passport()`	Process passport: RUN_ID, PID, build info
`get_runtime_counters()`	Runtime counters: calls, blocks, warnings
`intel_execution_timeline(limit)`	Recent action timeline with durations
`intel_get_project_context(root)`	Single snapshot: state, index, health, memory
`intel_explain_project_state(root)`	Human-readable project state diagnosis
`intel_tool_health()`	Tool success rates, latency, confidence
`refresh_db_connection()`	Reset database handle and reconnect

🏗️ Architecture

Clean Architecture with DI Container

┌──────────────────────────────────────────────────────────────────┐
│ MCP Server (~1000 lines) │
│ src/mcp/server.py + server_tools.py + server_factory.py │
│ │
│ ┌──────────────────────────────────────────────────────────┐ │
| │ DI Container (18 services) │ │
│ │ src/core/di_container.py — ServiceCollection │ │
│ │ │ │
│ │ ┌──────────┐ ┌────────────┐ ┌──────────────────────┐ │ │
│ │ │ Indexer │ │ Searcher │ │ DebounceBatch │ │ │
│ │ │ Embedder │ │ SymbolIdx │ │ CircuitBreaker │ │ │
│ │ │ Parser │ │ FileGuard │ │ RateLimiter │ │ │
│ │ └──────────┘ └────────────┘ └──────────────────────┘ │ │
│ └──────────────────────────────────────────────────────────┘ │
│ │ │
│ ┌────────────┴────────────┐ │
│ ▼ ▼ │
│ ┌────────────────────┐ ┌────────────────────────────────────┐ │
│ │ 28 Tool Classes │ │ 16 intel_* + 13 inline tools │ │
│ │ src/mcp/tools/*.py │ │ intelligence/layer.py + │ │
│ │ + codebase hub │ │ server_tools.py (inline) │ │
│ │ Constructor Inj. │ │ error_boundary decorator │
│ │ 1 execute_script │ │ asyncio.wait_for(timeout) │ │
│ └────────────────────┘ └────────────────────────────────────┘ │
└──────────────────────────────────────────────────────────────────┘
 │
 ▼
┌─────────────────┐ ┌───────────────────┐
│ RemoteEmbedder │ │ LanceDB v2 │
│ (llama.cpp GGUF │ │ (Vector DB) │
│ native, primary;│ │ BM25 + Vector │
│ ONNX INT8 │ │ │
│ in-process │ │ │
│ fallback) │ │ │
└─────────────────┘ └───────────────────┘

⚡ Performance

Mode	Latency	Best For
`search_code(query, mode="fast")`	~80-500ms	Simple keyword / exact name
`search_code(query, mode="quality")`	~250-2000ms	Semantic search with reranker
`search_code(query, mode="deep")`	~2-5s	Complex research across modules
`search_code(query, mode="context")`	~200-800ms	Find similar code by fragment
`get_symbol_info(query)`	~200-1500ms	Symbol definition + call graph
`impact_analysis(symbol)`	~1-5s	Change impact analysis

Environment Variables

Variable	Default	Description
`LM_STUDIO_HOST`	`localhost`	LM Studio hostname
`LM_STUDIO_PORT`	`1234`	LM Studio port
`OLLAMA_HOST`	`localhost`	Ollama hostname
`OLLAMA_PORT`	`11434`	Ollama port
`LOG_LEVEL`	`INFO`	Logging verbosity level
`MSCODEBASE_MCP_TOOLS`	(default set)	Comma-separated list of visible tools (e.g. `search_code,codebase`)
`MSCODEBASE_EXECUTE_SCRIPT_ENABLED`	`false`	Enable `execute_script` tool (RCE risk)
`LLAMA_BACKEND`	`auto`	Reranker backend: `auto` / `msvc` (CPU) / `vulkan` (GPU)
`MSCODEBASE_REMOTE_TOKEN`	(empty)	Bearer token for remote mode (`src/remote_main.py`, Streamable HTTP). Empty = auth disabled
`MSCODEBASE_REMOTE_RATE_LIMIT_RPS`	`30.0`	Remote gate rate limit (requests/sec per key: per-token + per-IP). `0` = disabled

EMBEDDING_MODEL (ранее в таблице) — больше не используется: модель определяется автоматически (llama.cpp GGUF, fallback ONNX e5-small INT8).

🔧 Troubleshooting

MCP Server Not Responding

Symptoms: tools timeout, no response.

Checklist:

File → Quit → reopen the project
Run python install.py to reconfigure
Check logs: %LOCALAPPDATA%\mscodebase\logs\ (data_root)

Index Empty (0 chunks)

Run in Agent Panel:

intel_trigger_reindex()

Then verify: codebase(action="index", path="status")

LM Studio Connection Issues

# Verify the server responds:
python -c "import urllib.request; print(urllib.request.urlopen('http://localhost:1234/v1/health').read())"

Expected: {"status":"ok"}.

📁 Project Structure

mscodebase-intelligence/
├── src/
│ ├── main.py # MCP server entry point (~194 lines)
│ ├── mcp/
│ │ ├── server.py # MCP server creation (~597 lines)
│ │ ├── server_factory.py # DI setup + server lifecycle (~478 lines)
│ │ ├── server_tools.py # Tool registration + 13 inline tools (~607 lines)
│ │ └── tools/ # 17 modules + base class
│ │ ├── codebase_tool.py # codebase(action=...) hub + execute_script
│ │ ├── search_tools.py # search_code, get_symbol_info, impact_analysis
│ │ ├── indexing_tools.py # notify_change, index_project_dir, index_health
│ │ ├── git_tools.py # get_branch_info, get_commit_history, get_file_history
│ │ ├── system_tools.py # get_index_status, get_health_report, read_live_file, get_logs
│ │ ├── analysis_tools.py # structural_search, get_repo_map, get_repo_rank, scan_changes
│ │ ├── graph_tools.py # cross_repo_search, cross_project_deps, graph_query
│ │ ├── investigation_tools.py # get_bug_correlation, get_hotspots, find_similar_bugs
│ │ ├── lifecycle_tools.py # submit_background_task, get_task_status, verify_action
│ │ ├── meta_tools.py # IndexTool, GitTool, SystemTool (spoke tools for codebase hub)
│ │ └── write_tools.py # WriteTool (rename, move, delete, replace, insert)
│ ├── core/ # Business logic + backward-compat shims
│ │ ├── di_container.py # ★ DI Container (18 services, ServiceCollection)
│ │ ├── error_handler.py # error_boundary decorator + ToolError
│ │ ├── rate_limiter.py # SlidingWindowRateLimiter + DebounceBatch + CircuitBreaker
│ │ ├── graph.py # PropertyGraph (29 edge types)
│ │ ├── structural_search.py # 13 AST patterns (Tree-sitter)
│ │ ├── lsp_client.py # Thin LSP client (pyright JSON-RPC 2.0)
│ │ ├── intelligence_layer.py # Shim → core/intelligence/layer.py
│ │ ├── indexing/ # 18 files: indexer, parser, symbol_index, file_guard, ...
│ │ ├── search/ # 18 files: engine (Searcher), scoring, bm25, cypher_*, ...
│ │ └── intelligence/ # 5 files: layer (intel_* tools), jobs, health, context, store
│ ├── providers/
│ │ ├── embedder/
│ │ │ └── remote_embedder.py # ONNX e5-small INT8 + LM Studio/Ollama fallback
│ │ └── reranker/ # llama_runner, multi_provider, search_result_reranker, scoring
│ ├── config/
│ │ └── settings.py # All configuration via os.getenv (Single Source of Truth)
│ └── utils/ # paths, i18n, ui_formatter, zed_config
├── docs/
│ ├── en/ # English docs
│ ├── ru/ # Russian docs
│ └── zh/ # Chinese docs
├── scripts/ # CLI utilities (install, sync, benchmark, audit)
├── tests/ # 853 tests (pytest)
├── install.py # Installer (3 languages: en/ru/zh)
└── README.md

🛠️ Development

See docs/en/CONTRIBUTING.md for:

How to add new MCP tools
Test structure and CI pipeline
Commit message conventions

Quick Start for Devs

# Setup
python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt
# Run MCP server directly (test)
python -m src.main
# Run tests
pytest tests/ -m "not integration and not benchmark"
# Live smoke (реальные сервисы, без моков — обязательно после изменений в серверах/индексе)
python scripts/smoke_e2e.py --project .
# Live smoke памяти (негативный контроль verify-on-read: VERIFIED/REFUTED/терминальные guard'ы)
python scripts/smoke_memory.py

📄 License

MIT License — see LICENSE for details.

🙏 Acknowledgments

Zed IDE — code editor
LM Studio — local LLM inference
LanceDB — vector database
Model Context Protocol — MCP standard

Env Access Extractor

Ported from codebase-memory-mcp (MIT, DeusData 2025). See src/core/indexing/parser.py (ENV_FUNCS_BY_LANG, ENV_MEMBERS_BY_LANG, _walk_env_accesses_iter) and tests/test_env_extractor.py (22 tests).

Name		Name	Last commit message	Last commit date
.github/workflows		.github/workflows
.local		.local
.locks		.locks
adapters		adapters
deploy/docker		deploy/docker
docs		docs
examples/plugins/verify_claim		examples/plugins/verify_claim
experiments		experiments
extensions/vscode/mscodebase-sync		extensions/vscode/mscodebase-sync
locales		locales
logo		logo
scripts		scripts
src		src
telemetry		telemetry
tests		tests
tools		tools
.dockerignore		.dockerignore
.env.example		.env.example
.gitignore		.gitignore
AGENTS.md		AGENTS.md
AGENT_DIARY.md		AGENT_DIARY.md
AI_INSTALLATION_PROMPT.md		AI_INSTALLATION_PROMPT.md
CODE_OF_CONDUCT.md		CODE_OF_CONDUCT.md
CONTRIBUTING.md		CONTRIBUTING.md
DEV_DIARY.md		DEV_DIARY.md
EXPERIMENTS_LOG.md		EXPERIMENTS_LOG.md
ISSUE.md		ISSUE.md
KNOWN_ISSUES.md		KNOWN_ISSUES.md
LICENSE		LICENSE
MANIFEST.in		MANIFEST.in
README.md		README.md
SECURITY.md		SECURITY.md
WISDOM.md		WISDOM.md
__mscodebase_ext__.marker		__mscodebase_ext__.marker
extension.toml		extension.toml
install.bat		install.bat
install.py		install.py
install.sh		install.sh
pyproject.toml		pyproject.toml
requirements-lock.txt		requirements-lock.txt
requirements.txt		requirements.txt
ruff.toml		ruff.toml
start_remote.bat		start_remote.bat
start_server.bat		start_server.bat
uninstall.bat		uninstall.bat

Latest commit

History

Folders and files

Repository files navigation

MSCodebase Intelligence

🎯 Positioning

What you get

LSP: Hybrid Rename Only

Platforms

Languages

✨ Features

🚀 Quick Start

Providers

📚 Documentation Map

Research & Writeups

🔧 MCP Tools (65 total)

Core Search

LSP Analysis (basedpyright)

Index Management (via codebase(action="index", ...))

System & Diagnostics

Analytics

Git & History (via codebase(action="git", ...))

Lifecycle & Verification

Write Tools — codebase(action=...)

Intelligence Layer (intel_*) — 16 High-Level Tools

Dev Tools (4)

Diagnostic Tools (7)

🏗️ Architecture

Clean Architecture with DI Container

⚡ Performance

Environment Variables

🔧 Troubleshooting

MCP Server Not Responding

Index Empty (0 chunks)

LM Studio Connection Issues

📁 Project Structure

🛠️ Development

Quick Start for Devs

📄 License

🙏 Acknowledgments

Env Access Extractor

About

Topics

Resources

Code of conduct

Contributing

Security policy

Stars

Watchers

Forks

Used by

Contributors

Languages

Index Management (via `codebase(action="index", ...)`)

Git & History (via `codebase(action="git", ...)`)

Write Tools — `codebase(action=...)`