mscodebase-intelligenceBLOCK
Intelligent codebase search & indexing for Zed. Async MCP server featuring LanceDB/BM25 hybrid search, multi-bucket RAG, and autonomous self-healing workflows. High-performance, memory-safe, and ready for your production code.
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
🇬🇧 English • 🇷🇺 Русский • 🇨🇳 中文
AI-powered semantic code search for Zed IDE — deep code analysis MCP server
[](https://www.python.org/downloads/) [](https://opensource.org/licenses/MIT) [](https://modelcontextprotocol.io/) [](https://zed.dev/) [](https://github.com/ManSio/mscodebase-intelligence/actions/workflows/ci.yml) [](tests/)
Features • Quick Start • Tools • Documentation • Installation • Architecture • Contributing • Security
Last updated: 2026-09-23
🎯 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) │ │ │ └───────────────────────────────────────────────┘ │ │ │ │ │
021505d68592OBSERVED · 2026-10-11Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add mscodebase-sync --env API_KEY=${API_KEY} --env FIRST_KEY=${FIRST_KEY} --env LLAMA_EMBED_MAX_TOKENS=${LLAMA_EMBED_MAX_TOKENS} --env LLAMA_RERANK_MAX_TOKENS=${LLAMA_RERANK_MAX_TOKENS} -- npx -y [email protected]{
"mcpServers": {
"mscodebase-sync": {
"command": "npx",
"args": [
"-y",
"[email protected]"
],
"env": {
"API_KEY": "${API_KEY}",
"FIRST_KEY": "${FIRST_KEY}",
"LLAMA_EMBED_MAX_TOKENS": "${LLAMA_EMBED_MAX_TOKENS}",
"LLAMA_RERANK_MAX_TOKENS": "${LLAMA_RERANK_MAX_TOKENS}"
}
}
}
}Exposed tools (14)
14 read · 0 write · 0 destructive.
| Tool | Risk | Description |
|---|---|---|
ack_impact | read | Подтверждает осведомлённость о влиянии изменений в файле. |
debug_runtime_passport | read | Диагностика: возвращает |
dual_arm_health_check | read | Двухрукая проверка здоровья (mutation + health). |
get_health_report | read | Полная диагностика системы: индекс, bridge, health, providers. |
get_logs | read | Возвращает последние ошибки и предупреждения из логов проекта. |
get_runtime_counters | read | Возвращает счётчики runtime: сколько запросов выполнено, |
intel_execution_timeline | read | Лента последних действий системы (имплиситный success signal). |
intel_explain_project_state | read | Человекочитаемый диагноз состояния проекта. |
intel_get_project_context | read | Единый снэпшот состояния проекта: state, index, bridge, health, |
intel_tool_health | read | Панель здоровья инструментов:成功率, latency, confidence, routes. |
notify_change | read | Обновляет индекс одного файла через LSP VFS или диск. |
ping | read | Детерминированный эхо-тул: проверяет эквивалентность stdio/HTTP на манер |
read_live_file | read | Читает файл из памяти LSP (включая несохранённые изменения). |
refresh_db_connection | read | Сбрасывает handle БД и переподключается к LanceDB. |
Trust audit
BLOCKgrade F · trust 26/100 Do not install this without reading the findings. The audit found something that could harm you or your machine.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | FAIL |
| L2 | Instruction surface (what it tells the agent) | FAIL |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (5 observation(s))
- Network
- declared (5 observation(s))
- Shell
- declared (15 observation(s))
- Dependencies
- not all pinned
- Secrets in source
- found
Findings (25)
"Модуль ловит только **известные формы** секретов по структурному префиксу (AKIA..., sk-ant-..., ghp_..., xox..., -----BEGIN PRIVATE KEY-----, URL с кредами, Bearer-токен и т.п.), а не по энтропии — ч
"Модуль ловит секреты по принципу **префикс-анкоренных структурных паттернов** (AKIA..., sk-ant..., ghp_..., `-----BEGIN PRIVATE KEY-----`, URL с кредами, Bearer и т.п.), а не по энтропии — чтобы не р
"Модуль ловит только **известные формы** секретов по структурному префиксу (AKIA..., sk-ant-..., ghp_..., -----BEGIN PRIVATE KEY-----, URL с кредами, Bearer-токены) — принцип prefix-anchored, а не ent
"Модуль ловит секреты **только по структурному префиксу** (AKIA..., sk-ant-..., ghp_..., -----BEGIN PRIVATE KEY-----, URL с кредами, `Bearer <токен>`, и присваивания вида `KEY=` лишь если значение дли
"Модуль **ненадёжен как граница безопасности** — он ловит только ИЗВЕСТНЫЕ ФОРМЫ по структурному префиксу; новый формат, секрет разбитый по строкам или записанный словами — проходит. Принцип: **prefix
"yaml.load(", "pickle.loads(", "pickle.load(",exec(compile(module, "<verifier>", "exec"), {})"eval(", "exec(", "compile(","__import__", "eval(", "exec(", "compile(",# WISDOM.md — дистиллят фактов (≤50 строк)
opencode.cmd :
opencode.cmd :
opencode.cmd :
opencode.cmd :
mod = __import__(mod_name)
mod = __import__(mod_name)
r"^exfiltrate",
sys.settrace(_trace)
sys.settrace(None)
print(f"\nЛожные отзывы по rename-якорям: {n_false}/{total_checked} = {rate:.1%}")print("\nВЕРДИКТ H3: отсутствие load-циклов = импорт не падает (runtime-safe).")print("\nПрогоны (чередование, min по 2):")print("\nПоказываю что РЕАЛЬНО получает LLM с TESTS-signal и без него.\n")print(f"\nИтого: {sum(1 for _, p in results if p)}/{len(results)} атак отражено", flush=True)<th style="border: 1px solid #ddd; padding: 8px; background-color: #4a90e2; color: white; text-align: left;">Signal</th>
Gates applied: critical_finding, no_behavioural_pass.
021505d68592full audit observations/trust-audit/mcp-server/mansio__mscodebase-intelligence.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-11 | 021505d68592 | BLOCK | F | 26 | first audit |
Questions
What is the mscodebase-intelligence MCP server?
Intelligent codebase search & indexing for Zed. Async MCP server featuring LanceDB/BM25 hybrid search, multi-bucket RAG, and autonomous self-healing workflows. High-performance, memory-safe, and ready for your production code.
What tools does mscodebase-intelligence expose?
14 in total: 14 read-only, 0 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.
Is mscodebase-intelligence safe to connect to an agent?
No — not without reading the findings first. The audit graded it F (26/100) and found 14 critical or high issues in the source. Each one is listed on this page with the file and line it is on.
What credentials does mscodebase-intelligence need?
It reads API_KEY, FIRST_KEY, LLAMA_EMBED_MAX_TOKENS, LLAMA_RERANK_MAX_TOKENS, MSCODEBASE_ACK_SECRET, MSCODEBASE_REMOTE_TOKEN, OPENROUTER_API_KEY, SECOND_KEY, TOKEN, TOKENIZERS_PARALLELISM and UNIQUE_KEY from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does mscodebase-intelligence run?
It speaks streamable-http, so it runs as a service you connect to over the network. It is published on npm as mscodebase-sync at 0.1.0.
How current is this page?
The grade is for one exact copy of the source (021505d68592), read on 2026-10-11. The repository is watched and re-audited when it changes.