Cortex MCP — Multi-Level Reasoning Server

io.github.j0hanz由 GitHub 账号 j0hanz 发布v1.11.0更新于 Oct 24, 2026

Multi-level reasoning MCP server with configurable depth levels

已验证STDIO仅桌面知识与记忆AI 与机器学习

概览

AI 生成的概览

本地 stdio 服务器,为助手提供有状态、可配置深度的多步推理工具,并保留可查看的会话轨迹。

可以做什么
  • 以 basic、normal、high 或 expert 深度运行并延续有步数上限的推理会话,可逐步进行或批量完成
  • 按步骤记录结构化的观察、假设与评估,回滚到较早步骤,或提前给出结论结束会话
  • 以资源形式读取活动会话列表、单个会话的 JSON 视图、完整 markdown 轨迹和单条思考
  • 使用内置提示按指定深度开始推理、继续会话、重试失败任务或获取帮助
适用场景
适合希望助手以明确、有上限的步骤推演难题,并留下可查看或审计轨迹,而不是一次性给出推理的场景。
运行要求
需要 Node.js 24 或更高版本,通过 npx 在本地运行该 npm 包,以及支持 stdio 传输的 MCP 客户端。Docker 仅用于容器镜像,可选。无需账号、API 密钥或请求头;可选环境变量可调整会话有效期、会话数量、令牌上限和轨迹脱敏。
安装前请注意
推理工具并非只读且不具幂等性:它会创建并修改内存中的会话,回滚会丢弃后续思考。会话在进程重启后丢失,默认 30 分钟后过期。除非启用轨迹脱敏,思考文本会原样保存,请勿写入机密信息。

安装

在 SourceWeft 中

  1. 打开 控制台中的 Cortex MCP — Multi-Level Reasoning Server,将其添加到工作区。
  2. 为需要使用其工具的对话启用该服务。

Desktop only,通过 STDIO。 STDIO 服务会启动本地进程,因此需要 SourceWeft 桌面宿主。

其他 MCP 客户端

参照 仓库 中的启动说明。

README

Cortex MCP

[npm version] [License]

[Install in VS Code] [Install in VS Code Insiders] [Install in Visual Studio]

[Add to LM Studio] [Install in Cursor] [Install in Goose]

Multi-level reasoning MCP server with configurable depth levels, session-based state management, structured thought input, and real-time trace resources.

Overview

Cortex MCP is a stdio-only MCP server for stateful, depth-controlled reasoning. The runtime entrypoint in src/index.ts connects createServer() to StdioServerTransport, and the server surface in src/server.ts enables tools, prompts, completions, logging, and subscribable resources around a single session-based reasoning engine.

The live MCP surface confirmed by Inspector is 1 tool, 6 concrete resources, 4 resource templates, and 7 prompts. Sessions are stored in memory, exposed as MCP resources, and cleared on process restart.

Key Features

  • reasoning_think supports step-by-step sessions, run_to_completion batches, rollback, early conclusion, and structured observation / hypothesis / evaluation input.
  • Four depth levels are built into the engine: basic, normal, high, and expert, each with bounded thought ranges and token budgets.
  • Prompt helpers expose reasoning.basic, reasoning.normal, reasoning.high, reasoning.expert, reasoning.continue, reasoning.retry, and get-help.
  • Resource endpoints expose internal docs plus live session lists, per-session JSON views, full markdown traces, and individual thought documents.
  • Completions are wired for levels, session IDs, and thought names through completable() and resource-template completion hooks.

Requirements

  • Node.js >=24 for local npx or npm usage.
  • An MCP client that supports stdio transport.
  • Optional: Docker if you want to build or run the container image defined by Dockerfile.

Quick Start

Use this standard MCP client configuration:

json
{  "mcpServers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

Client Configuration

Install in VS Code

[Install in VS Code]

Add to .vscode/mcp.json:

json
{  "servers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

Or install via CLI:

sh
code --add-mcp '{"name":"cortex-mcp","command":"npx","args":["-y","@j0hanz/cortex-mcp@latest"]}'

For more info, see VS Code MCP docs.

Install in VS Code Insiders

[Install in VS Code Insiders]

Add to .vscode/mcp.json:

json
{  "servers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

Or install via CLI:

sh
code-insiders --add-mcp '{"name":"cortex-mcp","command":"npx","args":["-y","@j0hanz/cortex-mcp@latest"]}'

For more info, see VS Code Insiders MCP docs.

Install in Cursor

[Install in Cursor]

Add to ~/.cursor/mcp.json:

json
{  "mcpServers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

For more info, see Cursor MCP docs.

Install in Visual Studio

[Install in Visual Studio]

Add to mcp.json (VS integrated):

json
{  "mcpServers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

For more info, see Visual Studio MCP docs.

Install in Goose

[Install in Goose]

Add to Goose extension registry:

json
{  "mcpServers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

For more info, see Goose MCP docs.

Install in LM Studio

[Add to LM Studio]

Add to LM Studio MCP config:

json
{  "mcpServers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

For more info, see LM Studio MCP docs.

Install in Claude Desktop

Add to claude_desktop_config.json:

json
{  "mcpServers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

For more info, see Claude Desktop MCP docs.

Install in Claude Code

Add to Claude Code CLI:

json
{  "mcpServers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

Or install via CLI:

sh
claude mcp add cortex-mcp -- npx -y @j0hanz/cortex-mcp@latest

For more info, see Claude Code MCP docs.

Install in Windsurf

Add to ~/.codeium/windsurf/mcp_config.json:

json
{  "mcpServers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

For more info, see Windsurf MCP docs.

Install in Amp

Add to Amp MCP config:

json
{  "mcpServers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

Or install via CLI:

sh
amp mcp add cortex-mcp -- npx -y @j0hanz/cortex-mcp@latest

For more info, see Amp MCP docs.

Install in Cline

Add to cline_mcp_settings.json:

json
{  "mcpServers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

For more info, see Cline MCP docs.

Install in Codex CLI

Add to ~/.codex/config.yaml or codex CLI:

json
{  "mcpServers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

For more info, see Codex CLI MCP docs.

Install in GitHub Copilot

Add to .vscode/mcp.json:

json
{  "servers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

For more info, see GitHub Copilot MCP docs.

Install in Warp

Add to Warp MCP config:

json
{  "mcpServers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

For more info, see Warp MCP docs.

Install in Kiro

Add to .kiro/settings/mcp.json:

json
{  "mcpServers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

For more info, see Kiro MCP docs.

Install in Gemini CLI

Add to ~/.gemini/settings.json:

json
{  "mcpServers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

For more info, see Gemini CLI MCP docs.

Install in Zed

Add to ~/.config/zed/settings.json:

json
{  "context_servers": {    "cortex-mcp": {      "settings": {        "command": "npx",        "args": ["-y", "@j0hanz/cortex-mcp@latest"]      }    }  }}

For more info, see Zed MCP docs.

Install in Augment

Add to VS Code settings.json:

Add to your VS Code settings.json under augment.advanced.

json
{  "augment.advanced": {    "mcpServers": [      {        "id": "cortex-mcp",        "command": "npx",        "args": ["-y", "@j0hanz/cortex-mcp@latest"]      }    ]  }}

For more info, see Augment MCP docs.

Install in Roo Code

Add to Roo Code MCP settings:

json
{  "mcpServers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

For more info, see Roo Code MCP docs.

Install in Kilo Code

Add to Kilo Code MCP settings:

json
{  "mcpServers": {    "cortex-mcp": {      "command": "npx",      "args": ["-y", "@j0hanz/cortex-mcp@latest"]    }  }}

For more info, see Kilo Code MCP docs.

Use Cases

Start bounded reasoning at the right depth

Use reasoning.basic, reasoning.normal, reasoning.high, or reasoning.expert when the client wants a prompt-first entrypoint, or call reasoning_think directly with query, level, and the first thought. Each response returns the current session state plus a summary string that tells the client how to continue.

Relevant tool: reasoning_think
Related prompts: reasoning.basic, reasoning.normal, reasoning.high, reasoning.expert

Continue, retry, or batch an active session

Reuse sessionId to continue a prior trace, switch to runMode="run_to_completion" when you already have the remaining thought inputs, or use the continuation and retry prompts to generate the next call payload. The handler also supports rollbackToStep and isConclusion for revising or ending a trace early.

Relevant tool: reasoning_think
Related prompts: reasoning.continue, reasoning.retry

Inspect live traces without re-running the tool

Read reasoning://sessions for the active session list, reasoning://sessions/{sessionId} for the JSON detail view, reasoning://sessions/{sessionId}/trace for the markdown transcript, or reasoning://sessions/{sessionId}/thoughts/{thoughtName} for a single thought. This lets a client present progress or audit a session independently from the next tool call.

Relevant resources: reasoning://sessions, reasoning://sessions/{sessionId}, reasoning://sessions/{sessionId}/trace, reasoning://sessions/{sessionId}/thoughts/{thoughtName}

Architecture

text
[MCP Client]    |    | stdio    v[src/index.ts]    createServer()    -> new StdioServerTransport()    -> server.connect(transport)    |    v[src/server.ts]    McpServer("cortex-mcp")    capabilities:      - tools      - prompts      - completions      - logging      - resources { subscribe: true, listChanged: true }    |    +--> tools/call    |     -> reasoning_think    |     -> src/tools/reasoning-think.ts    |     -> ReasoningThinkInputSchema / ReasoningThinkToolOutputSchema    |     -> src/engine/reasoner.ts    |     -> SessionStore    |    +--> prompts/get    |     -> src/prompts/index.ts    |    +--> resources/read    |     -> src/resources/index.ts    |     -> internal://* and reasoning://sessions/*    |    +--> notifications          -> logging messages          -> resources/list_changed          -> resources/updated          -> notifications/progress

Request Lifecycle

text
[Client] -- initialize --> [Server][Server] -- serverInfo + capabilities --> [Client][Client] -- notifications/initialized --> [Server][Client] -- tools/call {name: "reasoning_think", arguments} --> [Handler][Handler] -- validate args --> [Reasoner + SessionStore][Reasoner] -- progress/resource events --> [Server notifications][Handler] -- structuredContent + optional trace resource --> [Client]

MCP Surface

Tools

reasoning_think

Stateful reasoning tool for creating and continuing multi-step sessions. It supports one-step interactive calls, run_to_completion batches, structured observation/hypothesis/evaluation input, rollback, and early conclusion while returning structured session state.

ParameterTypeRequiredDescription
querystringnoQuestion or problem to analyze.
levelstringnoDepth level. Required for new sessions. basic (1–3 steps, 2K budget), normal (4–8 steps, 8K budget), high (10–15 steps, 32K budget), expert (20–25 steps, 128K budget).
targetThoughtsintegernoExact step count. Must fit level range.
sessionIdstringnoSession ID to continue.
runModestringno"step" (default) or "run_to_completion".
thoughtanynoReasoning text. Stored verbatim. String for step mode, string[] for batch.
isConclusionbooleannoEnd session early at final answer.
rollbackToStepintegerno0-based index to rollback to. Discards later thoughts.
stepSummarystringnoOne-sentence step summary.
observationstringnoKnown facts at this step.
hypothesisstringnoProposed next idea.
evaluationstringnoCritique of hypothesis.
Data Flow
text
1. [Client] -- tools/call {name: "reasoning_think", arguments} --> [Server]   Transport: stdio2. [Server] -- dispatch("reasoning_think") --> [Handler: src/tools/reasoning-think.ts]3. [Handler] -- validate(ReasoningThinkInputSchema) --> [src/engine/reasoner.ts]4. [Reasoner] -- create/update session --> [src/engine/session-store.ts]5. [Handler] -- structuredContent + optional embedded trace resource --> [Client]

Resources

ResourceURI or TemplateMIME TypeDescription
server-instructionsinternal://instructionstext/markdownUsage instructions for the MCP server.
server-configinternal://server-configapplication/jsonRuntime limits and level configurations for the reasoning server.
tool-cataloginternal://tool-catalogtext/markdownTool reference: models, params, outputs, data flow.
tool-infointernal://tool-info/{toolName}text/markdownPer-tool contract details.
tool-info-reasoning_thinkinternal://tool-info/reasoning_thinktext/markdownContract details for reasoning_think.
workflowsinternal://workflowstext/markdownRecommended workflows and tool sequences.
reasoning.sessionsreasoning://sessionsapplication/jsonList of active reasoning sessions with summaries. Updated in real-time as sessions progress.
reasoning.sessionreasoning://sessions/{sessionId}application/jsonDetailed view of a single reasoning session, including all thoughts and metadata.
reasoning.tracereasoning://sessions/{sessionId}/tracetext/markdownMarkdown trace of a reasoning session (full content).
reasoning.thoughtreasoning://sessions/{sessionId}/thoughts/{thoughtName}text/markdownMarkdown content of a single thought (for example Thought-1).

Prompts

PromptArgumentsDescription
get-helpnoneReturn server usage instructions.
reasoning.basicquery required, targetThoughts optionalBasic-depth reasoning (1-3 thoughts).
reasoning.normalquery required, targetThoughts optionalNormal-depth reasoning (4-8 thoughts).
reasoning.highquery required, targetThoughts optionalHigh-depth reasoning (10-15 thoughts).
reasoning.expertquery required, targetThoughts optionalExpert-depth reasoning (20-25 thoughts).
reasoning.continuesessionId required, query optional, level optionalContinue an existing session. Optional follow-up query.
reasoning.retryquery required, level required, targetThoughts optionalRetry a failed reasoning task with modified parameters.

MCP Capabilities

CapabilityStatusEvidence
toolsconfirmedsrc/server.ts:203-205, src/tools/reasoning-think.ts:479
promptsconfirmedsrc/server.ts:205, src/prompts/index.ts:201
completionsconfirmedsrc/server.ts:207, src/prompts/index.ts:249, src/resources/index.ts:375
loggingconfirmedsrc/server.ts:204, src/server.ts:98
resources.subscribeconfirmedsrc/server.ts:208, src/server.ts:121
resources.listChangedconfirmedsrc/server.ts:208, src/server.ts:114
progress notificationsconfirmedsrc/lib/mcp.ts:71, src/tools/reasoning-think.ts:424

Tool Annotations

AnnotationValueEvidence
readOnlyHintfalsesrc/tools/reasoning-think.ts:500
destructiveHintfalsesrc/tools/reasoning-think.ts:502
openWorldHintfalsesrc/tools/reasoning-think.ts:503
idempotentHintfalsesrc/tools/reasoning-think.ts:501

Structured Output

  • reasoning_think declares outputSchema and returns structuredContent, with an embedded trace resource when the trace is small enough. Evidence: src/tools/reasoning-think.ts:498, src/lib/mcp.ts:97-114.

Configuration

VariableDefaultRequiredEvidence
CORTEX_SESSION_TTL_MS1800000 (30 minutes)nosrc/engine/reasoner.ts:22, src/engine/session-store.ts:19
CORTEX_MAX_SESSIONS100nosrc/engine/reasoner.ts:23, src/engine/session-store.ts:20
CORTEX_MAX_TOTAL_TOKENS2000000nosrc/engine/reasoner.ts:24, src/engine/session-store.ts:21
CORTEX_MAX_ACTIVE_REASONING_TASKS32nosrc/engine/config.ts:41-44
CORTEX_REDACT_TRACE_CONTENTfalsenosrc/engine/config.ts:21

[!NOTE] The source does not define any HTTP host/port configuration. The only other environment-related signal is NODE_ENV=production in the Docker image and --env-file=.env in the local dev:run script.

Security

ControlStatusEvidence
input validationconfirmedsrc/schemas/inputs.ts:13, src/schemas/outputs.ts:46, src/prompts/index.ts:207
stdout-safe logging fallbackconfirmedsrc/server.ts:98, src/server.ts:145
main-thread-only runtimeconfirmedsrc/index.ts:15-25
non-root container userconfirmedDockerfile:37

[!NOTE] No auth, OAuth, HTTP origin checks, or rate-limiting controls are implemented in the current source because the server only exposes stdio transport.

Development

ScriptCommandPurpose
devtsc --watch --preserveWatchOutputWatch and compile source during development.
dev:runnode --env-file=.env --watch dist/index.jsRun the built server in watch mode with an optional local .env file.
buildnode scripts/tasks.mjs buildClean dist, compile TypeScript, copy assets, and make the entrypoint executable.
linteslint .Run ESLint across the repository.
type-checknode scripts/tasks.mjs type-checkRun source and test TypeScript checks concurrently.
testnode scripts/tasks.mjs testRun the TypeScript test suites with the configured loader.
test:distnode scripts/tasks.mjs test:distRebuild first, then run tests against the built output.
test:fastnode --test --import tsx/esm src/__tests__/**/*.test.ts node-tests/**/*.test.tsRun the fast direct test command without the task wrapper.
formatprettier --write .Format the repository.
inspectornpm run build && npx -y @modelcontextprotocol/inspector node dist/index.jsBuild the server and open it in the MCP Inspector.
prepublishOnlynpm run lint && npm run type-check && npm run buildEnforce release checks before publishing.

Additional helper scripts for diagnostics, coverage, asset copying, and knip are defined in package.json.

Build and Release

  • .github/workflows/release.yml bumps package.json and server.json, then runs npm run lint, npm run type-check, npm run test, and npm run build before tagging and creating a GitHub release.
  • The same workflow publishes the package to npm with Trusted Publishing, publishes to the MCP Registry with mcp-publisher, and pushes a multi-arch Docker image to ghcr.io.
  • Dockerfile uses a multi-stage Node 24 Alpine build, prunes dev dependencies, and runs the released container as the mcp user.

Troubleshooting

  • Sessions are in memory and expire after 30 minutes by default. If you receive E_SESSION_NOT_FOUND, start a new session or increase CORTEX_SESSION_TTL_MS.
  • runMode="run_to_completion" requires enough thought entries to cover the remaining steps. If you want the server to return after each step, keep the default step mode.
  • For stdio transport, do not add custom stdout logging around the server process. This server routes logs through MCP logging and falls back to stderr on failures.

Credits

DependencyRegistry
@modelcontextprotocol/sdknpm
zodnpm

Contributing and License

  • License: MIT
  • Contributions are welcome via pull requests.

来源:README.md,提交 c1171fd

工具

0
SourceWeft 尚未在沙箱中启动过该软件包,因此其工具未知。

版本历史

1
  1. v1.11.0最新Sep 16, 2026

评分

已满足适用检查项的 40%

该评分衡量此收录条目的完整程度,依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价,也不表示服务器安全或当前可用:可连通性是带日期的检查结果,评分也不包含安装量或用户评价。

  • 近 30 天内可连通25 分

    请保持端点能够响应 MCP 握手;SourceWeft 每 30 天探测一次。

  • 已声明认证方式10 分

    请在注册中心条目中声明凭据请求头,或保持端点能够响应探测,以便判定其认证方式。

  • 已列出工具15 分

    请让服务器在无凭据时响应 tools/list,或发布可在沙箱中无凭据启动的软件包。

  • 描述必需15 分
  • 图标10 分
  • 文档15 分
  • 许可证10 分

    请在仓库或注册中心条目中声明许可证。