be-amsa-reporter-te
Automates BE AMSA review by validating EFF folders, driving AMSA 5.1 to generate reports and graphs, and producing a PowerPoint summary. Designed to not hang with worker threads and timeouts.
README
be-amsa-reporter-te
MCP server that automates a BE AMSA review end-to-end. Trigger it in chat with:
create a backend amsa report
Refactored, optimized and watchdog-guarded successor to be-amsa-manager-te. It keeps only
the proven, active automation path, runs the whole multi-minute pipeline on a worker thread
behind a hard timeout (so the chat can never hang), and stages all EFF/report I/O on local
disk to avoid slow SMB round-trips.
What it does
- Lists the sub-folders under
\\Mucsdn31\ATV_Power_Pro\MOSFET_TE\Test_program_status_tracking\Copilot_BE_AMSA\effdataand asks which one to analyse. - Validates the chosen folder: exactly one
.efffile ending in each of_S1R1,_S1R2,_S2R1,_S2R2(no duplicates). If not, it returns an actionable message. - Stages the four EFFs locally and drives AMSA 5.1:
- Launches it from a clean slate (kills any stale
GageAnalysis.exe) and dismisses the Precondition failed popup with Yes. File > New→ Setup Dialog → adds the four EFF files withTestSetup= 1, 1, 2, 2.- Sets the option groups (Test ID by Name and Number, Device no further identification, Unit no rescaling + dB/log(A) linearisation, Override die coordinates).
Start→ ticks the listed Test Nr to exclude → Continue → Continue → OK.- Reads the results grid, sorts by GrR, self-calibrates the row geometry, and for every test with GrR ≥ 10 % ensures both S (Scatter) and P (Probability) are marked; unmarks any below threshold. Marks are verified and re-tried until they converge.
Configuration > Setup information→ sets each row's Description to its EFF file name (matched by Setup No / Rep No).Results > Write HTML-Report→ writes the report + aGraphs\sub-folder of JPGs.
- Launches it from a clean slate (kills any stale
- Copies the report + Graphs back to the share and builds a PowerPoint into
Copilot_BE_AMSA\generatedppt\<folder>.pptx:- Slide 1:
<folder> AMSA Review - Slide 2: Data Description (the four EFF file names)
- Slide 3: GrR Summary (
Gr&R > 25 %and10 % ≤ Gr&R ≤ 25 %) - Slide 4: GrR Summary (all tests) — bar chart with 10 %/25 % red boundary lines
- Slide 5+: one slide per test with GrR ≥ 10 %, scatter + probability JPG + assessment.
- Slide 1:
Why it will not hang
The create_backend_amsa_report tool only awaits the folder pick on the event loop; the
entire blocking pipeline runs on a worker thread (anyio.to_thread.run_sync). A watchdog
(BEAR_RUN_TIMEOUT, default 20 min) wraps the run — on timeout the worker is abandoned,
AMSA is force-killed and a clear timeout result is returned. A progress heartbeat
(BEAR_HEARTBEAT_SECONDS, default 10 s) reports the current phase so the run never looks
stuck.
Install
cd C:\Users\maunahan\.copilot\MCPservers\be-amsa-reporter-te
python3.11 -m pip install -r requirements.txt
Registration (mcp.json)
Registered in C:\Users\maunahan\AppData\Roaming\Code\User\mcp.json:
"be-amsa-reporter-te": {
"type": "stdio",
"command": "C:/Users/maunahan/AppData/Local/Microsoft/WindowsApps/python3.11.exe",
"args": ["-m", "be_amsa_reporter_te.server"],
"cwd": "C:/Users/maunahan/.copilot/MCPservers/be-amsa-reporter-te",
"env": {
"BEAR_BASE_ROOT": "\\\\Mucsdn31\\ATV_Power_Pro\\MOSFET_TE\\Test_program_status_tracking\\Copilot_BE_AMSA",
"BEAR_AMSA_LNK": "C:\\ProgramData\\Microsoft\\Windows\\Start Menu\\Programs\\AMSA 5.1\\AMSA 5.1.lnk"
}
}
Restart the MCP server (Command Palette → MCP: List Servers → be-amsa-reporter-te → Restart) after any code/config edit — a running server keeps the old modules in memory.
Configuration
Everything is overridable via BEAR_* environment variables (no code changes needed).
Common knobs: BEAR_MAIN_WINDOW_TITLE_RE, BEAR_UI_BACKEND, BEAR_EXCLUDE_TEST_NRS,
BEAR_GRR_PLOT_THRESHOLD, BEAR_GRR_HIGH_THRESHOLD, BEAR_RUN_TIMEOUT,
BEAR_HEARTBEAT_SECONDS, and the results-grid geometry / control IDs in config.py.
Modules
config.py— all paths/selectors/tunables fromBEAR_*env.folders.py— list + validate effdata folders.amsa_driver.py— pywinauto automation of AMSA 5.1 (active path only).html_report.py— parse the generated HTML for Test Nr / name / GrR / limits.ppt_report.py— build the review PowerPoint from the HTML +Graphs\JPGs.server.py— FastMCP server exposingcreate_backend_amsa_report, worker thread + watchdog + heartbeat.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。