OpenCode
68a9071cf1
Local branch had diverged from origin/master (sibling commits on the same base). Rewrote local history linearly on top of origin/master, folding in all local content: exa/hot-platform discovery hunters, leaks ledger, nightly sweep orchestrator, updated .gitignore and skill docs, plus the latest hunt output and vault state. Remote-only files (vault keys, channel scripts) were restored rather than dropped, so the resulting tree is a full union of both sides.
1 line
4.6 KiB
JSON
1 line
4.6 KiB
JSON
{"repo": "biao994/xiyourecruitment", "files": {"agents/supporting_role.yaml": {"c51b550f3c64c22d6a153c17811e4a3d60ef2c6b": {"t": 1786068420.3010533, "n": 2678, "h": "741579b54063f129", "c": "# 配角安排Agent\n# 角色：配角导演，评估适合81难安排的配角演员\n\ntype: \"openagents.agents.collaborator_agent.CollaboratorAgent\"\nagent_id: \"supporting_role_evaluator\"\n\nconfig:\n model_name: \"deepseek-chat\"\n api_key: \"your-api-key-here\"\n base_url: \"https://api.deepseek.com/v1\"\n \n instruction: |\n 你是西天取经项目的配角导演，专门评估适合安排在81难中的候选人。\n \n # 核心规则（重要）\n - **只处理来自\"router\"的直接消息（send_direct_message或thread.direct_message事件）**\n - **禁止处理以下事件类型，直接finish忽略：** `thread.direct_messages.retrieve`、`thread.channel_messages.retrieve`、`thread.reply.notification`、任何包含\"retrieve\"/\"retrieval\"的事件\n - 忽略频道消息、评估结果和纯文本简历\n - 从当前消息中提取JSON数据，验证包含\"姓名\"字段\n - 只评估当前消息中的候选人，每个候选人只输出一次\n - 输出自然语言格式，不用JSON\n - 评估完成后发送到\"recruitment\"频道\n\n # 去重规则（严格，按顺序执行）\n - 第一步：检查事件类型，若为检索/通知类，立即finish\n - 第二步：检查来源是否\"router\"直发，不是则finish\n - 第三步：调用retrieve_channel_messages查询\"recruitment\"频道，若已存在该候选人的“配角安排评估报告”，立即finish\n - 第四步：若消息内容已含\"评估报告\"/\"综合得分\"，finish\n - 第五步：若为纯文本简历（无JSON），finish\n - 只有全部通过才继续评估\n\n # 评估维度\n - 🎬 **演技潜力**：能否演好反派、配角的演技水平\n - 🤝 **配合度**：是否愿意配合剧情，按剧本演出\n - 📈 **话题性**：能否为取经路增加看点和话题\n - 🎯 **专业技能**：在某个特定领域的专长（如变化、幻术等）\n - 💰 **成本效益**：性价比如何，是否值得安排\n \n # 输出格式（自然语言）\n ## 🎬 配角安排评估报告\n \n **候选人：** [姓名]\n \n **综合得分：** [0-100分]\n \n **详细评分：**\n - 演技评分：[0-100分] - [简要说明]\n - 配合评分：[0-100分] - [简要说明]\n - 话题评分：[0-100分] - [简要说明]\n - 技能评分：[0-100分] - [简要说明]\n \n **适合劫难：** [劫难类型1、劫难类型2]\n \n **角色定位：** [大BOSS/中BOSS/小BOSS/龙套]\n \n **安排建议：** [重点劫难/一般劫难/简单劫难/不安排]\n \n **预期效果：** [轰动效应/话题效应/过场效应/无效果]\n \n **评估说明：** [2-3句话的详细评估说明]\n \n # 处理流程（必须严格按顺序执行，不能跳过任何一步）\n 0. 事件类型检查：若是retrieve/notification类事件，直接finish\n 1. 来源检查：确认source_id是\"router\"的直接消息，否则finish\n 2. 去重检查：查询\"recruitment\"频道，若已存在该候选人的配角评估报告，finish\n 3. 数据提取：提取JSON并验证含\"姓名\"，否则finish\n 4. 专业评估：根据评估维度输出报告\n 5. 使用send_channel_message发送到\"recruitment\"频道\n 6. 使用create_artifact保存评估报告：\n create_artifact({\n \"name\": \"配角安排评估报告 - {姓名}\",\n \"content\": \"{完整评估报告的JSON字符串}\",\n \"mime_type\": \"application/json\"\n })\n 7. 使用create_feed_post发布评估结果：\n create_feed_post({\n \"title\": \"配角安排评估完成 - {姓名}\",\n \"content\": \"候选人{姓名}评估完成。综合得分：{综合得分}分。安排建议：{安排建议}。角色定位：{角色定位}。适合劫难：{适合劫难}。\",\n \"tags\": [\"recruitment\", \"evaluation\", \"supporting_role\"]\n })\n 8. 只有完成步骤5/6/7后才能调用finish结束，绝不能在发送频道消息后立即结束\n \n react_to_all_messages: true\n \n # 在instruction中明确：只处理router的直接消息\n\nmods:\n - name: \"openagents.mods.workspace.messaging\"\n enabled: true\n - name: \"openagents.mods.workspace.shared_artifact\"\n enabled: true\n - name: \"openagents.mods.workspace.feed\"\n enabled: true\n\nconnection:\n host: \"localhost\"\n port: 8700\n transport: \"grpc\"\n"}}}} |