LLM 기반의 온디맨드 도구 생성 시스템으로, 인간의 승인 절차와 안전한 실행 환경을 제공합니다.
이 시스템은 LLM을 사용하여 TypeScript 도구를 동적으로 생성하며, 실행 전 인간의 승인을 요구하고 샌드박스 환경에서 실행합니다.
Related MCP server: code2mcp
┌─────────────┐ ┌──────────────┐ ┌─────────────┐
│ Synthesizer │────▶│ Approval │────▶│ Sandbox │
│ (LLM) │ │ (Human Gate) │ │ (Execution) │
└─────────────┘ └──────────────┘ └─────────────┘
│ │ │
▼ ▼ ▼
Generates TS Waits for Runs in
tool code human approval isolated env파일 | 목적 |
| OpenAI 호환 LLM을 사용하여 도구 코드 생성 |
| 인간 개입 게이트 — 실행 전 승인 필요 |
| 생성된 코드를 위한 안전한 실행 환경 |
| 도구 지속성 및 저장 |
| MCP 서버 통합 |
| 런타임 구성 관리 |
이 도구는 모든 OpenAI 호환 LLM API와 작동합니다:
OpenRouter — 100개 이상의 모델 (Claude, GPT, Llama 등)
OpenAI — GPT-4o, o3 등
Ollama — 로컬 모델 (Llama, Qwen 등)
LM Studio — GUI 기반 로컬 모델
Groq — 빠른 추론
기타 모든 OpenAI 호환 API
# Install dependencies
npm install
# Copy environment template
cp .env.example .env.env 파일을 편집하여 공급자 세부 정보를 입력하세요:
# Option 1: OpenRouter (default - 100+ models)
LLM_API_KEY=your-openrouter-key
LLM_BASE_URL=https://openrouter.ai/api/v1
LLM_MODEL=anthropic/claude-sonnet-4-6
# Option 2: OpenAI direct
LLM_API_KEY=sk-...
LLM_BASE_URL=https://api.openai.com/v1
LLM_MODEL=gpt-5.4
# Option 3: Ollama (local)
LLM_BASE_URL=http://localhost:11434/v1
LLM_MODEL=llama-3.3
# Option 4: Groq
LLM_API_KEY=gsk_...
LLM_BASE_URL=https://api.groq.com/openai/v1
LLM_MODEL=llama-3.3-70b-versatilenpm run build모든 것이 작동하는지 확인하는 가장 빠른 방법:
npx @modelcontextprotocol/inspector node dist/server.js이 서버는 모든 MCP 클라이언트와 작동합니다. 예시 구성:
Claude Desktop — Claude Desktop MCP 설정에 추가:
{
"mcpServers": {
"jit-tool-synthesis": {
"command": "node",
"args": ["/absolute/path/to/jit-tool-synthesis/dist/server.js"],
"env": {
"LLM_API_KEY": "your-api-key",
"LLM_BASE_URL": "https://openrouter.ai/api/v1",
"LLM_MODEL": "anthropic/claude-sonnet-4-6"
}
}
}
}Claude Code:
claude mcp add jit-tools node /absolute/path/to/jit-tool-synthesis/dist/server.jsVS Code (Copilot):
code --add-mcp '{"name":"jit-tools","type":"stdio","command":"node","args":["/absolute/path/to/jit-tool-synthesis/dist/server.js"]}'Cursor — .cursor/mcp.json에 추가:
{
"mcpServers": {
"jit-tools": {
"command": "node",
"args": ["/absolute/path/to/jit-tool-synthesis/dist/server.js"],
"env": { "LLM_API_KEY": "your-api-key" }
}
}
}재시작 없이 LLM 공급자를 변경할 수 있습니다:
# View current config
get_config
# Change model at runtime
set_config model=openai/gpt-5.4도구 | 설명 |
| 자연어로 새로운 도구 생성 |
| 승인 전 샘플 매개변수로 보류 중인 도구 테스트 |
| 보류 중인 도구 활성화 |
| 보류 중인 도구 폐기 |
| 승인된 도구 실행 |
| 승인된 모든 도구 목록 표시 |
| 도구 세부 정보 보기 |
| 도구 삭제 |
| 승인 대기 중인 도구 목록 표시 |
| LLM 구성 보기 |
| 런타임에 LLM 공급자/모델 변경 |
요청 — 사용자가 도구 생성 요청 (예: "색상 변환기 생성")
합성 — LLM이 도구 코드 생성
테스트 — 커밋 전 샘플 매개변수로 검증
승인 — 인간이 코드를 검토하고 승인
실행 — 샌드박스 환경에서 도구 실행
저장 — 승인된 도구는 세션 간 유지됨
변수 | 설명 | 기본값 |
| 공급자용 API 키 | (클라우드용 필수) |
| API 엔드포인트 | |
| 사용할 모델 | anthropic/claude-sonnet-4-6 |
또한 지원 (레거시): OPENROUTER_API_KEY, OPENAI_API_KEY, OPENAI_BASE_URL, SYNTHESIZER_MODEL
생성된 코드는 격리된 VM 샌드박스에서 실행됨
차단 패턴을 통해 위험한 코드(process, require, eval 등) 방지
API 키는 구성 파일에 저장되지 않음
프로덕션 준비 완료 — 1단계 완료.
| Name | Required | Description | Default |
|---|---|---|---|
| tool_name | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Tool has no description.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Tool has no description.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Tool has no description.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Tool has no description.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Tool has no description.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Tool has no description.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
| Name | Required | Description | Default |
|---|---|---|---|
| tool_name | Yes | ||
| params | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Tool has no description.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Tool has no description.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Tool has no description.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Tool has no description.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Tool has no description.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Tool has no description.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
| Name | Required | Description | Default |
|---|---|---|---|
| tool_name | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Tool has no description.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Tool has no description.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Tool has no description.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Tool has no description.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Tool has no description.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Tool has no description.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Returns a list of tools waiting for approval
| Name | Required | Description | Default |
|---|---|---|---|
No parameters | |||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full burden of behavioral disclosure. It states the action ('Returns a list') but doesn't cover critical aspects like permissions needed, pagination, rate limits, or what constitutes 'pending' status. This leaves significant gaps for a tool that likely interacts with an approval workflow.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, efficient sentence that directly states the tool's function without any fluff. It's front-loaded and wastes no words, making it easy for an agent to parse quickly.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the complexity of managing pending tools in an approval system, the description is insufficient. With no annotations, no output schema, and siblings like 'approve_tool' and 'reject_tool', it should explain more about the approval context, return format, or how 'pending' is defined. It leaves too much undefined for effective use.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The input schema has 0 parameters with 100% coverage, so no parameter documentation is needed. The description appropriately doesn't discuss parameters, earning a high baseline score for not adding unnecessary information.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the verb ('Returns') and resource ('list of tools waiting for approval'), making the purpose specific and understandable. However, it doesn't explicitly differentiate from sibling tools like 'list_tools' or 'get_tool', which might also retrieve tool information, so it falls short of a perfect score.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides no guidance on when to use this tool versus alternatives such as 'list_tools' or 'get_tool'. It lacks context about prerequisites, timing, or exclusions, leaving the agent to infer usage from the name alone.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Returns a list of all approved and available tools
| Name | Required | Description | Default |
|---|---|---|---|
No parameters | |||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full burden of behavioral disclosure. It states the tool returns a list but doesn't specify format, pagination, sorting, or any constraints like rate limits or authentication needs. For a tool with zero annotation coverage, this is a significant gap in transparency.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, efficient sentence that directly states the tool's purpose without any wasted words. It is front-loaded and appropriately sized for a simple tool with no parameters.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the lack of annotations and output schema, the description is incomplete. It doesn't explain what 'approved and available' means, the return format, or how this tool interacts with siblings. For a tool in a context with multiple related tools, more detail is needed to ensure proper use.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The tool has 0 parameters, and schema description coverage is 100%, so the schema already documents this fully. The description doesn't need to add parameter details, and it appropriately doesn't mention any. Baseline for 0 parameters is 4, as it avoids unnecessary information.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the verb ('Returns') and resource ('a list of all approved and available tools'), making the purpose understandable. However, it doesn't explicitly distinguish this tool from sibling tools like 'get_tool' or 'list_pending', which might have overlapping or related functionality.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides no guidance on when to use this tool versus alternatives like 'get_tool' (for a single tool) or 'list_pending' (for pending tools). It lacks explicit context, prerequisites, or exclusions, leaving usage unclear in relation to siblings.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
| Name | Required | Description | Default |
|---|---|---|---|
| tool_name | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Tool has no description.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Tool has no description.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Tool has no description.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Tool has no description.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Tool has no description.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Tool has no description.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
| Name | Required | Description | Default |
|---|---|---|---|
| tool_name | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Tool has no description.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Tool has no description.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Tool has no description.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Tool has no description.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Tool has no description.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Tool has no description.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Describe a capability you need and this will generate a working tool. Returns a pending tool that needs approval before it becomes available.
| Name | Required | Description | Default |
|---|---|---|---|
| description | Yes | What the tool should do | |
| example_input | No | Example input to guide generation | |
| example_output | No | Expected output format |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full burden. It discloses that the tool generates a pending tool requiring approval, which is useful behavioral context. However, it lacks details on permissions needed, rate limits, error conditions, or what 'pending' entails operationally (e.g., storage, visibility). For a mutation tool with zero annotation coverage, this is a significant gap in transparency.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is two sentences that are front-loaded and efficient: the first states the action and input, the second specifies the output and approval requirement. Every sentence earns its place with no wasted words, making it highly concise and well-structured.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's complexity (generating new tools with 3 parameters), lack of annotations, and no output schema, the description is minimally adequate. It covers the core purpose and approval workflow but misses details like error handling, return format, or integration with siblings. It's complete enough for basic understanding but has clear gaps for effective agent use.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so the schema already documents all three parameters ('description', 'example_input', 'example_output') with their purposes. The description doesn't add any parameter-specific semantics beyond what the schema provides, such as formatting tips or interdependencies. Baseline 3 is appropriate when the schema does the heavy lifting.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's purpose: 'generate a working tool' based on a described capability, with the specific outcome of returning 'a pending tool that needs approval'. It uses specific verbs ('generate', 'returns') and identifies the resource ('tool'), but doesn't explicitly differentiate from siblings like 'approve_tool' or 'execute_tool' in terms of workflow role.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies usage context by stating it 'generates a working tool' and that the result 'needs approval before it becomes available', which suggests it's part of a tool creation workflow. However, it doesn't explicitly state when to use this versus alternatives like 'approve_tool' or 'execute_tool', nor does it provide clear exclusions or prerequisites for use.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v2.0.0approve_toolexecute_toolget_toollist_pendinglist_toolsreject_toolremove_toolsynthesize_toolScored across 8 tools
Most tools have distinct purposes: list_pending and list_tools are clearly different (pending vs. all tools), synthesize_tool creates new tools, and approve_tool/reject_tool/remove_tool handle tool lifecycle. However, get_tool lacks a description, making its distinction from list_tools ambiguous—it might fetch a single tool's details, but this is unclear without context.
All tool names follow a consistent verb_noun pattern with snake_case, such as approve_tool, execute_tool, get_tool, list_pending, list_tools, reject_tool, remove_tool, and synthesize_tool. This uniformity makes the set predictable and easy to understand at a glance.
With 8 tools, this server is well-scoped for managing a tool synthesis and approval workflow. The count supports core operations like listing, creating, approving, rejecting, removing, and executing tools, without being overly sparse or bloated.
The tool set covers a complete lifecycle for tool synthesis: create (synthesize_tool), list (list_pending, list_tools), approve/reject (approve_tool, reject_tool), remove (remove_tool), and execute (execute_tool). A minor gap is the lack of descriptions for tools like get_tool, which might hinder full coverage if it's meant for detailed retrieval, but the core workflow is well-supported.
Runtime permission, approval, and audit layer for AI agent tool execution.
Give any MCP-compatible AI assistant a builder for live, hosted web tools and workflows.
The marketplace where agents don't just use tools — they build, publish, and compose new ones.
MCP server for your apps' tools and custom tools, plus hosted AI agents and approval-gated workflows
