{"$schema":"https://wellknown.network/schemas/agent-record-v1.json","schemaVersion":"1","id":"ag_g733npka2uxt","handle":"msprof-mcp","url":"https://wellknown.network/agents/msprof-mcp","links":{"self":"https://wellknown.network/agents/msprof-mcp/record.json","html":"https://wellknown.network/agents/msprof-mcp","markdown":"https://wellknown.network/agents/msprof-mcp/record.md","api":"https://wellknown.network/api/v1/agents/msprof-mcp","status":"https://wellknown.network/api/v1/agents/msprof-mcp/status","claim":"https://wellknown.network/agents/msprof-mcp/claim","claimApi":"https://wellknown.network/api/v1/claims","claimDescriptor":"https://wellknown.network/agents/msprof-mcp/claim.json","badge":"https://wellknown.network/agents/msprof-mcp/badge.svg","openapi":"https://wellknown.network/openapi.json"},"ard":{"identifier":"urn:air::server:msprof-mcp","type":"application/mcp-server-card+json"},"kind":"mcp_server","declared":{"name":"msprof-mcp","summary":"MCP server for Ascend Profiler (msprof) analysis","description":"# msprof mcp\n\n## 简介\nmsprof mcp 是一个基于 Model Context Protocol (MCP) 的服务器，旨在为大语言模型 (LLM) 提供分析 Ascend PyTorch Profiler 采集性能数据的能力。通过一系列内置工具，它可以帮助用户快速定位性能瓶颈、分析算子耗时、查看通信开销以及进行 Trace 数据的深度查询。\n\n## 目录结构\n```\nmsprof_mcp/\n├── pyproject.toml            # 项目配置文件 (build-system, dependencies)\n├── src/\n│   └── msprof_mcp/\n│       ├── __init__.py\n│       ├── server.py                 # MCP 服务器入口\n│       └── tools/                    # 工具包\n│           ├── msprof_analyze_cmd.py\n│           ├── csv_analyze.py\n│           ├── json_analyze.py\n│           └── trace_view/\n└── README.md\n```\n\n## MCP 能力说明\n\n本服务提供以下核心能力，支持多维度性能数据分析。您可以直接在对话中使用自然语言（如示例 Prompt）来调用这些工具。\n\n### 1. 总体分析 （msprof-analyze）\n\n| 工具名称 | 描述 | 示例 Prompt |\n| :--- | :--- | :--- |\n| `msprof_analyze_advisor` | 调用 `msprof-analyze advisor` 提供全方位性能建议（计算/调度瓶颈）。 | \"分析 `/path/to/data` 目录下的性能数据，找出主要瓶颈。\" |\n\n### 2. TimeLine 分析 (trace_view)\n\n| 工具名称 | 描述 | 示例 Prompt |\n| :--- | :--- | :--- |\n| `analyze_overlap` | 分析计算、通信与调度的重叠情况，判断负载特征（计算/通信密集型）。 | \"分析 `/path/to/trace_view.json` 的计算和通信重叠情况。\" |\n| `find_slices` | 搜索 Trace 中的特定 Slice（算子/函数），支持模糊匹配和时间范围过滤。 | \"在 `/path/to/trace_view.json` 中查找所有 'MatMul' 算子。\" |\n| `get_flow_data` | 根据时间范围获取 Flow 关联的 CPU/NPU 算子明细，支持按 `cpu_op` 或 `npu_op` 入口查询；结果过大时可通过 `result_output_path` 导出 CSV。 | \"获取 `/path/to/trace_view.json` 中 1000000000 到 2000000000 时间范围内的 NPU 算子关联 Flow 数据，并导出到 `/tmp/flow_data.csv`。\" |\n| `execute_sql_query` | 执行自定义 SQL 查询，支持 Slice/Thread/Process 等表的深度分析。 | \"对 `/path/to/trace_view.json` 执行 SQL 查询，统计耗时超过 1ms 的 Slice 数量。\" |\n\n### 3. 算子性能分析 (CSV)\n\n| 工具名称 | 描述 | 示例 Prompt |\n| :--- | :--- | :--- |\n| `analyze_kernel_details` | 分析 `kernel_details.csv`，提供耗时分布、Top N 算子、设备分布等。 | \"分析 `/path/to/kernel_details.csv`，列出耗时最长的 10 个算子。\" |\n| `get_operator_details` | 查询特定算子（按名称或类型）的详细执行信息。 | \"从 `/path/to/kernel_details.csv` 中获取 'FlashAttention' 算子的详细信息。\" |\n| `analyze_op_statistic` | 分析 `op_statistic.csv`，提供调用次数、总耗时及 Core 类型分布。 | \"统计 `/path/to/op_statistic.csv` 中的算子调用次数和总耗时。\" |\n| `get_op_type_details` | 查询特定类型算子或 Core 类…","publisher":null,"homepage":null,"repository":null,"version":"0.1.8","license":null,"protocols":["mcp"],"tags":["mcp"],"pricing":null,"endpoints":[{"url":"pypi:msprof-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":"data.database","name":"Databases","confidence":0.733,"provenance":"derived"}],"categories":["data"],"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":"msprof-mcp","registry":"pypi","observedAt":"2026-09-10T07:21:37.520Z","publishedAt":"2026-06-04T13:07:38.072088Z","latestVersion":"0.1.8"}},"verification":{"claimed":false,"claimedAt":null,"proofs":[]},"provenance":{"sources":[{"source":"pypi","key":"msprof-mcp","url":"https://pypi.org/project/msprof-mcp/","firstSeenAt":"2026-09-10T07:20:28.895Z","fetchedAt":"2026-09-10T07:20:28.895Z","normalizedAt":"2026-09-10T07:20:28.895Z"}]},"firstSeenAt":"2026-09-10T07:20:28.895Z","updatedAt":"2026-09-10T07:21:37.520Z"}