agent-report
从 Claude 代理 JSONL 文件中提取并显示最终消息。用于用户希望查看代理报告、查看代理输出、提取代理结果、检查代理产生的结果或读取子代理的最终响应时。触发于“显示我代理报告”、“代理 X 产生了什么”、“提取代理输出”、“查看代理结果”或“从代理 ad42ecb 获取报告”等请求。
📖 中文导读:从 Claude 代理 JSONL 文件中提取并显示最终消息。用于用户希望查看代理报告、查看代理输出、提取代理结果、检查代理产生的结果或读取子代理的最终响应时。触发于“显示我代理报告”、“代理 X 产生了什么”、“提取代理输出”、“查看代理结果”或“从代理 ad42ecb 获取报告”等请求。 以下正文为该技能的原始 SKILL.md 内容(英文),上方安装命令可直接使用。
name: agent-report description: Extract and display the final message from a Claude agent JSONL file. Use when the user wants to see an agent report, view agent output, extract agent results, check what an agent produced, or read the final response from a subagent. Triggers on requests like "show me agent report", "what did agent X produce", "extract agent output", "view agent results", or "get the report from agent ad42ecb".
Agent Report Extractor
Extract the final message from a Claude agent file and write it to a formatted markdown report.
Usage
Run the script with an agent ID:
scripts/extract_agent_message.sh <agent_id> [output_file]
agent_id: The short agent ID (e.g.,ad42ecb)output_file: Optional. Defaults toagent-<id>-report.mdin current directory
Workflow
- If the user provides an agent ID, run the script immediately
- If no agent ID is provided, ask the user for the agent ID using AskUserQuestion
- Run the script and confirm the output file location
Output Format
The script produces a markdown file containing:
- Agent slug/name
- Agent ID
- Model used
- Timestamp
- Full text content of the final message