Skills Plugins MCP Prompt Model 博客 我的中心
Development #image #mcp

smalltalk-mcp

Interact with live Smalltalk images (Squeak, Cuis) via MCP. Evaluate code, browse classes, view method source, define and modify classes and methods, query hierarchies and categories in a running Smalltalk environment.

DeepseekModel Curated skill Quality Good · 64 v1.0.0

Get

https://deepseekmodel.com/api/download.php?id=corporatesmalltalkconsultingltd-claudesmalltalk-skill-md&format=skill
Download .skill Standard format with system_prompt and model_config, ready for any agent framework
The actual content of the system_prompt field in the .skill file.
name smalltalk-mcp description Interact with live Smalltalk images (Squeak, Cuis) via MCP. Evaluate code, browse classes, view method source, define and modify classes and methods, query hierarchies and categories in a running Smalltalk environment. Smalltalk MCP Skill This skill connects Claude to a live Smalltalk image (Cuis or Squeak) via MCP. Setup (if tools are not yet connected) If the Smalltalk MCP tools are not available, help the user configure them: Prerequisites Python 3.10+ A Smalltalk VM: Cuis or Squeak The ClaudeSmalltalk repository: git clone https://github.com/CorporateSmalltalkConsultingLtd/ClaudeSmalltalk.git Build a ClaudeCuis.image following CUIS-SETUP.md. For Squeak, see SQUEAK-SETUP.md. Step 1: Install Python dependency pip install httpx If using Anthropic as the agent LLM provider, also: pip install anthropic Step 2: Create smalltalk-mcp.json Create this file in the ClaudeSmalltalk repo directory. All paths must be absolute. Example using Anthropic (copy from examples/smalltalk-mcp-anthropic.json ): { "version" : "1.0" , "model" : { "provider" : "anthropic" , "name" : "claude-sonnet-4-6" , "maxTokens" : 256000 , "apiKeyEnv" : "ANTHROPIC_API_KEY" } , "vm" : { "squeak" : "/absolute/path/to/Squeak6.0.app/Contents/MacOS/Squeak" , "cuis" : "/absolute/path/to/CuisVM.app/Contents/MacOS/Squeak" } , "image" : { "selected" : "cuis" , "squeak" : "/absolute/path/to/ClaudeSqueak.image" , "cuis" : "/absolute/path/to/ClaudeSmalltalk/ClaudeCuis.image" } , "transport" : { "type" : "stdio" , "args" : [ "--mcp" ] , "timeout" : 180 } } The user must set their API key: export ANTHROPIC_API_KEY=sk-ant-... Other provider examples are in the examples/ folder (Ollama, OpenAI, xAI, MQTT). Step 3: Configure Claude Desktop The user must edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS): { "mcpServers" : { "smalltalkAgent" : { "command" : "python3" , "args" : [ "/absolute/path/to/ClaudeSmalltalk/smalltalk_agent_mcp.py" ] , "env" : { "SMALLTALK_MCP_CONFIG" : "/absolute/path/to/ClaudeSmalltalk/.smalltalk-mcp.json" } } } } An example is at examples/claude_desktop_config.json . All paths must be absolute. After saving, Claude Desktop will reload and the 13 Smalltalk tools will become available. How to use the tools Once connected, you have 13 MCP tools for the live Smalltalk image. When to use smalltalk_task vs individual tools Use smalltalk_task for complex, multi-step work: "Review the Random class" — the agent browses, reads methods, and produces an assessment "Audit the Set class for correctness" "Define a Counter class with increment/decrement methods and tests" "Compare OrderedCollection and Array implementations" smalltalk_task delegates to a separate LLM configured in .smalltalk-mcp.json . You provide a natural language task and get back a complete result. This is the preferred tool for anything requiring multiple browse/evaluate steps. Use individual tools for quick, single operations: smalltalk_evaluate — run code: 3 factorial , Date today smalltalk_browse — get class metadata (superclass, ivars, method lists) smalltalk_method_source — read one method's source code smalltalk_list_classes — find classes by prefix smalltalk_hierarchy / smalltalk_subclasses — explore inheritance Best practices Always browse before modifying. Before defining or changing a method, use smalltalk_browse to understand the class structure and smalltalk_method_source to read existing implementations. Class-side methods. Use the side parameter with value "class" when viewing or defining class-side methods. The smalltalk_browse tool returns both instance and class methods. Class definitions. Use standard Smalltalk class definition syntax: Object subclass: #MyClass instanceVariableNames: 'foo bar' classVariableNames: '' poolDictionaries: '' category: 'MyCategory' Method source format. Provide complete method source including the selector line: increment count := (count ifNil: [0]) + 1. ^ count Testing. After defining methods, verify with smalltalk_evaluate : MyClass new increment Run SUnit tests: MyClassTest buildSuite run Exploring the system. Start broad, then narrow: smalltalk_list_categories — see what's in the image smalltalk_classes_in_category — explore a category smalltalk_browse — understand a class smalltalk_method_source — read specific methods Tool reference Tool Description smalltalk_task Run a complex task via the agent loop (preferred for multi-step work) smalltalk_evaluate Execute Smalltalk code and return the result smalltalk_browse Get class metadata: superclass, ivars, instance and class methods smalltalk_method_source View source code of a method (use side: "class" for class side) smalltalk_define_class Create or modify a class definition smalltalk_define_method Add or update a method on a class smalltalk_delete_method Remove a method from a class smalltalk_delete_class Remove a class from the system smalltalk_list_classes List classes matching a prefix smalltalk_hierarchy Get superclass chain for a class smalltalk_subclasses Get immediate subclasses of a class smalltalk_list_categories List all system categories smalltalk_classes_in_category List classes in a category
Keywords that activate this skill. Click one to copy it.

This skill does not provide trigger words.

The downloaded .skill package contains the following fields.
Field Description
formatFormat tag (skill/v1)
skill_idUnique skill ID
nameSkill name
versionVersion
descriptionDescription
categoryCategories (array)
trigger_wordsTrigger words
tagsTags
sourceSource
source_urlSource URL (this page)
exported_atExported at (set per download)
system_promptSystem prompt body
model_configModel config: provider / model / temperature / max_tokens / top_p
examplesExamples
install_guideImport guide for Coze / Dify / Claude / custom frameworks
The same skill can be exported in different platform formats.
.skill Standard format with system_prompt and model_config, ready for any agent framework Download
.skillpro Enhanced format with scripts, tools, dependencies and hooks Download
.json Plain JSON export with system_prompt and model parameters only Download
Coze Markdown with frontmatter, for Coze platform import Download
Dify Dify DSL, import directly after creating an app Download

每日精选 Skill 推荐,免费送到你邮箱

输入邮箱,每天接收一个精选 AI Agent 技能推荐。完全免费,持续更新。

验证码 --

提交后我们会发送一封确认邮件,点击邮件里的链接才会开始收信。

完全免费,取消任意时间。我们不会发送垃圾邮件。