{"$schema":"https://wellknown.network/schemas/agent-record-v1.json","schemaVersion":"1","id":"ag_hv6g6sfhgsh7","handle":"automodel-tau3-airline-mcp","url":"https://wellknown.network/agents/automodel-tau3-airline-mcp","links":{"self":"https://wellknown.network/agents/automodel-tau3-airline-mcp/record.json","html":"https://wellknown.network/agents/automodel-tau3-airline-mcp","markdown":"https://wellknown.network/agents/automodel-tau3-airline-mcp/record.md","api":"https://wellknown.network/api/v1/agents/automodel-tau3-airline-mcp","status":"https://wellknown.network/api/v1/agents/automodel-tau3-airline-mcp/status","claim":"https://wellknown.network/agents/automodel-tau3-airline-mcp/claim","claimApi":"https://wellknown.network/api/v1/claims","claimDescriptor":"https://wellknown.network/agents/automodel-tau3-airline-mcp/claim.json","badge":"https://wellknown.network/agents/automodel-tau3-airline-mcp/badge.svg","openapi":"https://wellknown.network/openapi.json"},"ard":{"identifier":"urn:air::server:automodel-tau3-airline-mcp","type":"application/mcp-server-card+json"},"kind":"mcp_server","declared":{"name":"automodel-tau3-airline-mcp","summary":"MCP server exposing the 14 official tau3-bench airline agent tools","description":"# τ³-bench airline MCP\n\n该包把固定版本 τ³-bench 的 14 个 airline Agent 工具暴露为标准 MCP；用户侧模拟器工具、评分条件、reset 和 snapshot 不会出现在 Agent 工具列表中。\n\n## 本地验证\n\n```bash\nuvx --from ./tools/tau3_airline_mcp tau3-airline-mcp --list-tools\nuvx --from ./tools/tau3_airline_mcp tau3-airline-mcp\n```\n\n第一条命令应输出 14 个工具及函数签名。第二条启动 stdio MCP，供 MCP Inspector 或本地客户端连接。\n\n## 打包\n\n```bash\nuv build tools/tau3_airline_mcp --out-dir dist/tau3_airline_mcp\n```\n\n产物为 `automodel_tau3_airline_mcp-0.1.0-py3-none-any.whl`。包内包含固定提交所需的 airline DB、policy、tasks、split 和最小 airline 运行代码，部署时不需要再从 GitHub 拉取完整 τ³-bench 框架。\n\n## 在百炼脚本部署\n\n百炼云端不能读取本机路径。官方文档给出的 Python 标准分发路径是公共 PyPI；也可先把 wheel 上传到可公开下载的 OSS 地址，或把本项目对应目录推送到公开 Git 仓库，然后：\n\n1. 打开百炼 **MCP 管理 → 创建 MCP 服务 → 使用脚本部署**。\n2. 安装方式选择 `uvx`，部署地域选择北京；低频评测选择基础模式。模板已经指定阿里云 PyPI 镜像，只用于安装 MCP SDK 与 Pydantic。\n3. 使用以下任一配置模板：\n   - `bailian-mcp-config.pypi.template.json`：发布到公共 PyPI 后使用，兼容性最明确；\n   - `bailian-mcp-config.wheel.template.json`：替换公开 OSS 地址，版本和内容容易冻结；\n   - `bailian-mcp-config.git.template.json`：直接从公开 Git 提交安装。\n4. 提交部署，在“工具”页确认恰好出现 14 个工具，并测试 `list_all_airports`。\n5. 在 airline Agent 中只挂载这个 MCP，粘贴 `agent_system_prompt_zh.md`，并在其后附上 `datasets/v1/resources/tau3/airline_policy.md` 完整原文。\n6. 将相同 Agent 配置复制为 AUTO、经济、均衡、性能四个版本，只修改模型模式。\n\n百炼官方支持使用 `uvx` 托管 Python stdio MCP，也支持连接远程 `streamableHttp` MCP。控制台入口和配置格式见[自定义 MCP 服务](https://help.aliyun.com/zh/model-studio/custom-mcp)。\n\n## 正式批量评测的状态隔离\n\n脚本部署版适合确认安装、14 个工具 schema 和单任务调用。它在一个 MCP 进程内维护 airline 数据库，百炼是否为每个 Agent 会话重建该进程并没有稳定契约，因此不能据此假设不同任务天然隔离。\n\n正式运行 `tau3_airline.jsonl` 时使用 `tools/tau3_openapi_adapter/`：\n\n- Agent 只看到 `/tools/*` 对应的 14 个工具；\n- 评测调度器在每个 `case_id` 前调用隐藏的 `/admin/reset`；\n- 任务结束后调用隐藏的 `/admin/snapshot`，交给官方 evaluator 判分；\n- 通过 AI 网关将 14 个业务接口导入为一个 MCP 服务，admin 接口不导入；\n- 禁止并发复用同一状态实例。需要并发时按 `run_id` 分片实例，或为服务增加按运行 ID 隔离的状态存储。\n\n这条远程路径多一个部署步骤，但才能保证端到端任务成功率可复现。","publisher":null,"homepage":null,"repository":null,"version":"0.1.0","license":null,"protocols":["mcp"],"tags":["mcp"],"pricing":null,"endpoints":[{"url":"pypi:automodel-tau3-airline-mcp","type":"package_pypi","auth":null,"probeable":false}],"skills":null,"tools":null,"extra":null,"attribution":{"kind":"pypi","name":"pypi","summary":"pypi","version":"pypi","description":"pypi"}},"derived":{"capabilities":[{"slug":"ai.prompting","name":"Prompt Management","confidence":0.802,"provenance":"derived"},{"slug":"dev.version-control","name":"Version Control","confidence":0.745,"provenance":"derived"}],"categories":["ai","dev"],"language":"en"},"observed":{"status":"unknown","statusReason":"Distributed as a package to run locally; no network endpoint to check.","lastOkAt":null,"lastProbedAt":null,"statusComputedAt":null,"reliability30d":null,"latestObservations":[],"tools":null,"package":{"name":"automodel-tau3-airline-mcp","registry":"pypi","observedAt":"2026-09-09T09:25:20.645Z","publishedAt":"2026-08-26T12:21:59.686515Z","latestVersion":"0.1.0"}},"verification":{"claimed":false,"claimedAt":null,"proofs":[]},"provenance":{"sources":[{"source":"pypi","key":"automodel-tau3-airline-mcp","url":"https://pypi.org/project/automodel-tau3-airline-mcp/","firstSeenAt":"2026-09-09T09:24:30.961Z","fetchedAt":"2026-09-09T09:24:30.961Z","normalizedAt":"2026-09-09T09:24:30.961Z"}]},"firstSeenAt":"2026-09-09T09:24:30.961Z","updatedAt":"2026-09-09T09:25:20.645Z"}