Define consistent strict and diagnostic validation for client list APIs

未關閉
#1,997 0 則留言 0 個 reaction 已指派 0 人 在 GitHub 檢視

維護者通常 1 天內回覆

還沒有人認領這個 Issue。

評估

難度
5/5
預估耗時
一週以上
新手友好度
35/100
Issue 類型
功能
描述清晰度
基本清楚
活躍度
活躍
技術堆疊
csharp

研究方向

Start by reading related issues #1996, #1993, and #1645 to understand the aggregation boundary, prior Skills APIs, and page-level cache hints. Then inspect the client list APIs for tools, prompts, resources, resource templates, and Skills, along with their tests and source-generated serialization paths. Define the shared validation and diagnostic behavior while preserving strict defaults, pagination and fatal page errors; add the mixed-page, later-page, rejection-position, and failure tests listed in the issue.

由索引模型根據 Issue 內容生成。

描述

P3

Summary

Evaluate a consistent validation contract for client list APIs for tools, prompts, resources, resource templates, and Skills, including an optional diagnostic path that exposes usable entries alongside explicit rejected-entry details.

Preserve existing strict defaults. This is separate from how many pages or entries a caller permits the SDK to fetch.

Motivation

A single malformed entry can prevent a host from using otherwise valid entries in a listing. Hosts may want to inspect individual failures, but silently dropping invalid entries would make an incomplete result look complete.

Skills previously had dedicated ListSkillsWithDiagnosticsAsync, SkillListingResult, and SkillEntryRejection APIs. #1993 removes that Skills-only design so equivalent concepts can be evaluated consistently across the SDK.

Proposed scope

  • Document the shared validation contract and any primitive-specific checks. Passing SDK structural checks must not be described as establishing trust or, for Skills, verifying file contents/frontmatter.
  • Evaluate an explicit opt-in diagnostic API or mode that returns accepted entries and rejection details, including original entry positions and actionable errors. Do not silently omit rejected entries.
  • Preserve page boundaries, continuation cursors, and page-level cache hints where appropriate. Do not invent a single aggregate TTL.
  • Distinguish recoverable individual-entry failures from malformed JSON, invalid envelopes, or missing/null required list members where a page cannot safely be recovered. Do not turn these into success-shaped empty results.
  • Preserve ordinary RPC failure and cancellation behavior, and support the SDK's source-generated/Native AOT serialization paths.
  • Test mixed valid/invalid pages, failures on later pages, rejection positions, fatal page errors, strict defaults, and the opted-in diagnostic behavior.

Boundaries

Do not bundle aggregation budgets, caching policy, or server-side pagination infrastructure into this issue. Existing single-page APIs remain the escape hatch for application-owned policy.

Coordinate with the aggregation-budget design on entry counting and incomplete-result reporting, while keeping the two policies independent.

Related work

  • Companion: #1996, caller-controlled aggregation budgets. That issue addresses when to stop fetching; this issue addresses what to do with malformed entries.
  • #1993 motivates replacing the Skills-specific partial-result helpers with a consistent cross-SDK design.
  • #1645 records the page-level caching-hint decisions that any new result shape should preserve.

[!NOTE]
This issue was drafted with GitHub Copilot.

主要語言
C#
星號
4.6k
分支
849
平均合併
1 天 16 小時
30 天內合併 PR
24

環境準備

在 Codespaces 中開啟

在瀏覽器裡用你自己的 GitHub 帳號啟動這個專案的開發容器。

  • 沒有 Dockerfile 或 Docker Compose 檔案
  • 沒有 Pull Request 範本
  • 閱讀貢獻指南

從這裡開始

  1. 先讀完整個 Issue,再讀專案的貢獻指南。
  2. 在 Issue 下留言說明你要接手 —— 這能避免兩個人做同樣的事。
  3. Fork 儲存庫,在一個分支上完成修改。
  4. 送出 Pull Request,並在描述裡引用這個 Issue 編號。

modelcontextprotocol/csharp-sdk 的其他 Issue

查看 modelcontextprotocol/csharp-sdk 的全部 Issue

相似的 Issue

更多 C# Issue

把新 issue 寄到你的電子郵件信箱

精選適合新手參與的 GitHub issue 摘要。