The server registers 22 tools with varying quality. Most tools have descriptions and input schemas present, but there are significant gaps in schema completeness, parameter descriptions, and error guidance. Many parameters lack type information or descriptions. Output schemas are not documented. The tool descriptions are present but often generic, missing context about when to use each tool versus alternatives. Several tools have complex parameters with enum constraints that are well-defined, but others have vague optional parameters. Error handling is minimal, most tools would fail silently or return cryptic API errors without recovery guidance. The composition is reasonable (tools are mostly single-responsibility), but parameter naming could be more explicit about expected types.
マガジンに記事を追加する
Obsidian MarkdownをnoteのHTML形式に変換(画像はプレースホルダーに置換)
サークル情報を取得する
記事へのコメント一覧を取得する
記事のスキ一覧を取得する
マガジンの詳細情報を取得する
メンバーシップの記事一覧を取得する
Most tool descriptions are too short (under 50 chars) and generic. Examples: 'マガジンの詳細情報を取得する' (22 chars), 'サークル情報を取得する' (12 chars), '記事へのコメント一覧を取得する' (17 chars). These lack context for LLM selection, they do not explain WHEN to use this tool vs similar tools, WHAT it returns beyond the obvious, or PREREQUISITES. Baseline target: 50-200 chars per pattern:tool-description.
No documented output schemas for any tool. The rubric requires tools to document what fields are returned so LLMs can plan downstream calls and extract the right data. E.g., does get-note return {id, title, body, author, created_at, likes_count}? What is the type of each field? Without this, agents cannot reliably chain tools or validate responses.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 54 | 2026-07-28+ | v2 |
| 2026-03-09 | D | 56 | - | v1 |
自分のメンバーシッププラン一覧を取得する
加入済みメンバーシップ一覧を取得する
記事の詳細情報を取得する
Get detailed information about a Notion page
テスト用:メンバーシップの記事一覧をダミーデータで取得する
テスト用:加入済みメンバーシップ一覧をダミーデータで取得する
Import a Notion page to note.com as a draft
List pages from a Notion database
下書き状態の記事を新規作成または更新する
画像付きの下書き記事を作成する(Playwrightなし、API経由で画像を本文に挿入)
Obsidian記事をnote下書き用に準備(画像情報を含む)
Preview how a Notion page will be converted to note.com format
マガジンから記事を削除する
note.comに画像をアップロード(アイキャッチ画像用。本文画像は publish-from-obsidian を使用)
note.comに複数の画像を一括アップロード
Parameter descriptions are missing or too vague on several tools. Examples: upload-image param 'imagePath' has description but missing guidance on supported formats, max size constraints (though code enforces 10MB max), or resolution limits. upload-images-batch 'imagePaths' array has no description of max array length, what happens if one fails, or whether partial success is possible.
No pagination guidance in tool descriptions. get-membership-notes accepts page/perPage params with defaults (page=1, perPage=20), but the description does not explain how pagination works, whether there is a next_cursor field returned, or what the maximum perPage value is. Baseline pattern:paginated-result requires tools returning lists to document pagination behavior and total count.
Mutually exclusive parameters not documented. upload-image has three alternative params (imagePath, imageUrl, imageBase64) but the description does not explicitly state they are mutually exclusive or what happens if multiple are provided. Code throws an error only if ALL are missing. LLMs may try to pass multiple, wasting a call.
No error recovery guidance. Tools do not document what can go wrong, how errors are classified (retryable vs fatal), or what the agent should do next. E.g., if upload-image fails with 'invalid MIME type', should the agent retry, convert the image, or ask the user? If get-note returns 404, should it search for the note first? No guidance.
Destructive operations (remove-magazine-note) lack confirmation or dry-run support. This tool will delete a note from a magazine without warning. pattern:confirmation-request recommends irreversible operations support a dry-run or confirmation step to prevent catastrophic mistakes.
No enum constraints on categorical parameters. post-draft-note-with-images has a 'bodyFormat' parameter with enum=['markdown', 'html'], which is well-defined. But other tools lack enums where they would help: e.g., no enum for membership status values, no enum for note visibility (private/public), no validation of magazine IDs or note IDs formats.
Test tools (get-test-membership-summaries, get-test-membership-notes) are exposed in production. These are mock/dummy data tools that should be hidden or gated behind a feature flag. They pollute the tool namespace, confuse users, and risk LLMs selecting test data when real data is intended.
post-draft-note-with-images 'eyecatch' parameter uses a union type (string | object), but the schema describes it as accepting both a filename string AND a Base64 image object. The description is helpful, but the schema structure is complex. LLMs may struggle to infer when to use which form. Simplify by offering separate parameters (eyecatch_filename vs eyecatch_base64) or better union documentation.