@mukundakatta/agentvet 用のMCPサーバー。 Claude Desktop、Cursor、Cline、Windsurf、Zed、その他あらゆるMCPクライアントで、LLMが生成したツール呼び出し引数を実行前に検証し、問題がある場合にLLMフレンドリーなリトライメッセージを生成できるようにします。
npx -y @mukundakatta/agentvet-mcp3つのツールを提供します：
validate_tool_args — 小さな形状仕様に基づいて引数をチェックします。{ valid, error?, retry_hint? } を返します。ここで retry_hint は、そのままLLMに送信できるフィードバックメッセージです。
lint_tool_definition — LLMのツール使用精度を低下させる一般的なミスがないか、ツール定義を健全性チェックします。
generate_retry_message — 検証エラーが与えられた場合、agentvetの ToolArgError.toLLMFeedback() フォーマットを使用して、LLM向けの標準的なリトライメッセージを作成します。
~/Library/Application Support/Claude/claude_desktop_config.json (macOS) または %APPDATA%\Claude\claude_desktop_config.json (Windows) を編集します：
{
"mcpServers": {
"agentvet": {
"command": "npx",
"args": ["-y", "@mukundakatta/agentvet-mcp"]
}
}
}Cursor (~/.cursor/mcp.json)、Cline、Windsurf、Zedでも同様の形式です。
Related MCP server: Echo MCP
validate_tool_args:
{
"tool_name": "send_email",
"args": { "to": "a@b.com" },
"shape": { "to": "string", "subject": "string", "body": "string" }
}戻り値:
{
"valid": false,
"error": "missing required field: subject",
"retry_hint": "send_email rejected your args: missing required field: subject. Please call again with the corrected arguments."
}lint_tool_definition:
{
"tool": {
"name": "BadName",
"inputSchema": { "type": "object", "properties": { "x": { "type": "string" } } }
}
}非snake_caseの名前、説明の欠如、フィールド説明の欠如、必須フィールドがないことなどに関する警告を返します。
generate_retry_message:
{
"tool_name": "send_email",
"validation_error": "missing required field: subject",
"attempted_args": { "to": "a@b.com" }
}ランタイムの呼び出し元が確認する標準的なリトライフィードバック文字列を返します。これにより、ライブエージェントループの外でリトライテキストを準備できます。
@mukundakatta/agentvet は依存関係のないJavaScriptライブラリです。このMCPサーバーは、その検証プリミティブをMCP対応のあらゆるAIアシスタントから利用可能にします。ツールレジストリを迅速に監査したり、チャットを離れることなく「この引数オブジェクトは私の send_email ツールにとって有効か？」とアシスタントに尋ねたりするのに便利です。
エージェントループ内でのランタイム引数検証には、Nodeプロセス内で直接 @mukundakatta/agentvet を使用してください（ツール関数をラップし、同期的に ToolArgError をスローします）。
agent-stackシリーズの一部です：
@mukundakatta/agentfit-mcp — Fit it. (適合させる)
@mukundakatta/agentguard-mcp — Sandbox it. (サンドボックス化する)
@mukundakatta/agentsnap-mcp — Test it. (テストする)
@mukundakatta/agentvet-mcp — Vet it. (審査する) (これ)
@mukundakatta/agentcast-mcp — Validate it. (検証する)
MIT
Given a tool name, validation error, and attempted args, build the canonical LLM-facing retry feedback message. Uses agentvet's ToolArgError.toLLMFeedback() formatting so the wording matches what runtime callers see.
| Name | Required | Description | Default |
|---|---|---|---|
| tool_name | Yes | ||
| validation_error | Yes | ||
| attempted_args | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries full burden. It describes inputs and output but does not disclose behavioral traits like idempotency, side effects, or read-only nature. The tool appears to be a pure function, but this is not explicitly stated. Minimal transparency beyond functional 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?
Two sentences that are concise and front-loaded. Every sentence provides essential information without redundancy. The structure is clear and efficient.
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?
For a simple string generation tool with 3 parameters and no output schema, the description covers inputs and the output's purpose. However, it does not describe the output format or provide examples, which would be helpful for an agent. Given the tool's simplicity and lack of annotations, the description could be more complete.
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 coverage is 0% (no parameter descriptions in schema). The description names the three parameters (tool_name, validation_error, attempted_args) and indicates their roles, but does not elaborate on expected formats, constraints, or examples. It adds value over the raw schema but does not fully compensate for the lack of schema descriptions.
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 specifies the tool's purpose: building a canonical LLM-facing retry feedback message given tool_name, validation_error, and attempted_args. It includes the formatting method (agentvet's ToolArgError.toLLMFeedback()), and distinguishes from siblings (lint_tool_definition, validate_tool_args) by its unique output.
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 when to use this tool—when generating a retry message after a validation error. It notes that the wording matches runtime callers, but does not explicitly state when not to use it or compare to alternatives. However, the siblings are sufficiently different, so no confusion.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Sanity-check a tool definition for common mistakes that hurt LLM tool-use accuracy: missing description, vague description, no required fields, schema fields without descriptions, non-snake_case names.
| Name | Required | Description | Default |
|---|---|---|---|
| tool | Yes | A tool definition: { name, description, inputSchema }. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations provided, so description carries full burden. It describes the analysis but does not explicitly state it is non-destructive or disclose output format. The behavior is implied but not fully transparent.
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, well-structured sentence with no unnecessary words. It efficiently communicates the tool's purpose and checks.
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?
No output schema is provided, and the description does not explain the return value (e.g., a list of issues). For a linting tool, this is a gap that reduces completeness for an AI agent.
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 coverage is 100% and the parameter 'tool' has a description explaining its structure. The description adds context about what to include, aiding correct invocation.
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 specifies the verb 'Sanity-check' and the resource 'tool definition', listing specific checks. It distinguishes from sibling tools (generate_retry_message, validate_tool_args) by its focus on definition quality.
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 for checking tool definitions, but lacks explicit when-to-use or when-not-to-use guidance. However, the context of sibling tools helps differentiate usage.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Validate a tool-call args object against a small shape spec. Returns { valid, error?, retry_hint? } where retry_hint is a ready-to-send LLM feedback message describing exactly what was wrong.
| Name | Required | Description | Default |
|---|---|---|---|
| tool_name | Yes | Name of the tool being called (surfaces in retry_hint). | |
| args | Yes | The args object the LLM wants to pass. | |
| shape | Yes | Shape spec mapping field name to type. Types: "string", "number", "boolean", "array", "object". Suffix with "?" for optional. Example: { "name": "string", "age": "number", "tags": "array", "notes": "string?" } |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Despite no annotations, the description transparently states the return structure ({valid, error?, retry_hint?}) and explains that retry_hint is a ready-to-send LLM feedback message. This covers the behavior well, though it omits details like idempotency or side effects.
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?
A single sentence with a clear structure: first states the action, then lists the return fields. It is concise but could be improved by front-loading the most critical information (e.g., validation outcome).
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?
The description covers the basic function and return values, but lacks usage context like when to validate or how the shape spec works (already in schema). No output schema means description should be more detailed, but it adequately explains the return format.
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 coverage is 100% with detailed descriptions for each parameter. The description adds minimal extra meaning beyond the schema (e.g., clarifying retry_hint's purpose), meeting the baseline for high coverage.
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 uses a specific verb ('Validate') and resource ('tool-call args object against a small shape spec'), clearly distinguishing it from sibling tools (generate_retry_message, lint_tool_definition) which serve different purposes.
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?
No explicit guidance on when to use this tool versus alternatives. The description implies usage for validating tool-call arguments but does not provide prerequisites or contrasting scenarios with siblings.
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.
v0.1.0generate_retry_messagelint_tool_definitionvalidate_tool_argsScored across 3 tools
Each tool targets a distinct operation: generating retry messages, linting tool definitions, and validating tool args. There is no overlap in purpose.
All tool names follow a consistent verb_noun snake_case pattern (generate_retry_message, lint_tool_definition, validate_tool_args), making them predictable.
With 3 tools, the server is well-scoped for its purpose of agent vetting and validation. Each tool earns its place without being too sparse or bloated.
The set covers core validation, linting, and feedback generation. A minor gap is the absence of a tool to suggest fixes or re-lint after changes, but the core workflow is complete.
MCP server for your apps' tools and custom tools, plus hosted AI agents and approval-gated workflows
MCP server providing access to the Scorecard API to evaluate and optimize LLM systems.
MCP server for progressive tool usage at any scale (see https://klavis.ai)
Nifty's MCP server — exposes tasks, projects, messages, and files as tools for AI agents.
