github / github/copilot-cli

Expose Skill/MCP Configuration Errors to Agent and Enable Agent Self-Awareness and Corrective Guidance

オープン
#3,646 コメント 0 件 リアクション 0 件 担当者 0 名 GitHub で見る
area:configuration area:mcp area:plugins
主要言語
Shell
スター
11.2k
フォーク
1.9k
平均マージ
14時間 16分
マージ済み PR(30日)
6

説明

### Describe the feature or problem you'd like to solve

Skill/MCP configuration errors should be exposed to the agent so that it can become aware of the issue and provide users with appropriate correction suggestions.

### Proposed solution

**Required Behavior:**

- When a user runs the skill command and discovers that some or all skills fail to load, the error messages shown to the user (such as malformed YAML header) should also be exposed to the Copilot agent. When the user asks for modifications, the agent should be aware of the existing issues.

- When a user runs the mcp command and finds that some or all MCP configurations fail to load, the error messages displayed to the user (such as JSON formatting errors) should likewise be exposed to the Copilot agent. When the user requests changes, the agent should clearly understand where the problem lies.

- The Copilot agent should understand its own configuration format instead of searching for documentation online when responding to user issues. In practice, the agent often retrieves incorrect documentation and modifies the configuration using the wrong format, resulting in even more inconsistent and problematic outcomes.

### Example prompts or workflows

Sample prompts:
- I ran `skill` and got: "Malformed YAML header". Can you fix my skill configuration?
- The `skill list` command shows that 2 out of 5 skills failed to load. Error: "Invalid field 'descriptin' in metadata." Please help me correct it.
- I ran `mcp` and it says: "JSON parse error: Unexpected token } in position 245." Can you fix my json configuration?

### Additional context

_No response_

コントリビューションガイド

コントリビューションガイドを開く

調査の方向性

Start by tracing the skill and mcp command flows and where their configuration errors are reported. Determine how those existing messages can reach the Copilot agent, then verify that failed skill or MCP configurations are identified accurately and that the agent provides corrective guidance using the documented formats.

索引モデルが issue の本文から書いたものです。

評価

技術スタック
shell
領域
ai, cli
issue の種類
機能追加
難易度
4/5
見積もり時間
3〜5日
活発さ
静か
明瞭さ
おおむね明確
初心者へのやさしさ
38/100

新しい issue をメールで受け取る

初心者向けの GitHub issue を短くまとめたダイジェスト。