AlphaMao1/AlphaMao_Skills

帮我安装这个 skill：https://github.com/AlphaMao1/AlphaMao_Skills/tree/main/skills/creator-research-tracker

帮我安装这个 skill：https://github.com/AlphaMao1/AlphaMao_Skills/tree/main/skills/progressive-investment-research

python scripts/scaffold_creator_dossier.py D:\research\creator-serenity --source-id serenity --display-name Serenity --platform x --locator https://x.com/aleabitoreddit --handle-or-channel aleabitoreddit

<http://xhslink.com/o/7cunSYmYq2d>

- `serenity`：X-first 来源。种子定位：`@aleabitoreddit` / `https://x.com/aleabitoreddit`。所有观点先按未验证 source lead 处理。

- `rihardjarc`：X + newsletter 来源。种子定位：`@RihardJarc` / `https://x.com/RihardJarc`，补充来源为 UncoverAlpha。适合 AI infrastructure、cloud CapEx、TPU/GPU economics、hyperscaler ASIC 和半导体供应链 source leads。账号拼写是 `RihardJarc`，不是 `RichardJarc`。

python scripts/scaffold_creator_dossier.py D:\research\creator-serenity --source-id serenity --display-name Serenity --platform x --locator https://x.com/aleabitoreddit --handle-or-channel aleabitoreddit

{"canonical_url":"https://x.com/aleabitoreddit/status/1000000000000000001","posted_at":"2026-06-08T02:10:00+00:00","author_handle":"aleabitoreddit","text":"NVDA supply chain bottleneck is shifting from accelerators to data center power and 

{"canonical_url":"https://x.com/aleabitoreddit/status/1000000000000000002","posted_at":"2026-06-08T03:20:00+00:00","author_handle":"aleabitoreddit","text":"AAOI is interesting but the real question is whether optical demand converts into su

{"canonical_url":"https://x.com/aleabitoreddit/status/1000000000000000003","posted_at":"2026-06-08T04:30:00+00:00","author_handle":"aleabitoreddit","text":"lol market is bored today"}

"id": "youtube_demo_fixture_1",       "title": "AI infrastructure bottlenecks and compute scaling",       "url": "https://www.youtube.com/watch?v=youtube_demo_fixture_1",

"title": "AI infrastructure bottlenecks and compute scaling",       "url": "https://www.youtube.com/watch?v=youtube_demo_fixture_1",       "channel": "YouTube Demo Source"

"url": "https://www.youtube.com/watch?v=youtube_demo_fixture_1",

"results": [     {       "url": "https://www.youtube.com/watch?v=youtube_demo_fixture_1",

{       "url": "https://www.youtube.com/watch?v=youtube_demo_fixture_1",       "ok": true,

"url": "https://www.youtube.com/watch?v=youtube_demo_fixture_1",

platforms:   - platform: x     locator: https://x.com/aleabitoreddit

- platform: x     locator: https://x.com/aleabitoreddit     handle_or_channel: aleabitoreddit

locator: https://x.com/aleabitoreddit

platforms:   - platform: x     locator: https://x.com/aleabitoreddit

- platform: x     locator: https://x.com/aleabitoreddit     handle_or_channel: aleabitoreddit

locator: https://x.com/aleabitoreddit

platforms:   - platform: x     locator: https://x.com/RihardJarc

- platform: x     locator: https://x.com/RihardJarc     handle_or_channel: RihardJarc

locator: https://x.com/RihardJarc

collection_method: chrome-plugin-first   - platform: newsletter     locator: https://www.uncoveralpha.com/

- platform: newsletter     locator: https://www.uncoveralpha.com/     handle_or_channel: UncoverAlpha

locator: https://www.uncoveralpha.com/

#!/usr/bin/env python3

Executable file

#!/usr/bin/env python3

DEFAULT_PROXY = "http://localhost:3456"

url = proxy_url.rstrip("/") + path     data = body.encode("utf-8") if body is not None else None     req = urllib.request.Request(url, data=data, method=method)

data = body.encode("utf-8") if body is not None else None     req = urllib.request.Request(url, data=data, method=method)     if body is not None:

req = urllib.request.Request(url, data=data, method=method)

req.add_header("Content-Type", "text/plain; charset=utf-8")     try:         with urllib.request.urlopen(req, timeout=timeout) as resp:

try:         with urllib.request.urlopen(req, timeout=timeout) as resp:             raw = resp.read().decode("utf-8")

with urllib.request.urlopen(req, timeout=timeout) as resp:

ensure_bridge(args.proxy_url)     handle = args.handle.lstrip("@")     locator = args.locator or f"https://x.com/{handle}"

handle = args.handle.lstrip("@")     locator = args.locator or f"https://x.com/{handle}"     target_id = open_timeline(args.proxy_url, locator)

locator = args.locator or f"https://x.com/{handle}"

#!/usr/bin/env python3

if not video_id:             continue         url = entry.get("url") or f"https://www.youtube.com/watch?v={video_id}"

continue         url = entry.get("url") or f"https://www.youtube.com/watch?v={video_id}"         if not str(url).startswith("http"):

url = entry.get("url") or f"https://www.youtube.com/watch?v={video_id}"

if not str(url).startswith("http"):             url = f"https://www.youtube.com/watch?v={video_id}"         videos.append(

url = f"https://www.youtube.com/watch?v={video_id}"

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

parser.add_argument("--lang", default="en.*", help="yt-dlp 字幕语言选择器，如 en.*、zh-Hans。")     parser.add_argument("--cookies-from-browser", help="给 yt-dlp 使用的浏览器 cookies，如 firefox 或 chrome。")     parser.add_argument("--proxy", help="代理 URL，如 htt

[Credential-related source omitted]

[Credential-related source omitted]

parser.add_argument("--proxy", help="代理 URL，如 http://127.0.0.1:7890。")

#!/usr/bin/env python3

#!/usr/bin/env python3

#!/usr/bin/env python3

#!/usr/bin/env python3

"x",                     "--locator",                     "https://x.com/aleabitoreddit",

"--locator",                     "https://x.com/aleabitoreddit",                     "--handle-or-channel",

"https://x.com/aleabitoreddit",

"x",                     "--locator",                     "https://x.com/aleabitoreddit",

"--locator",                     "https://x.com/aleabitoreddit",                     "--handle-or-channel",

"https://x.com/aleabitoreddit",

json.dumps(                     {                         "canonical_url": "https://x.com/aleabitoreddit/status/2063869376542192013",

{                         "canonical_url": "https://x.com/aleabitoreddit/status/2063869376542192013",                         "posted_at": "2026-06-08T06:22:58.000Z",

"canonical_url": "https://x.com/aleabitoreddit/status/2063869376542192013",

json.dumps(                     {                         "canonical_url": "https://x.com/aleabitoreddit/status/999999999999999999",

{                         "canonical_url": "https://x.com/aleabitoreddit/status/999999999999999999",                         "posted_at": "2026-06-08T05:00:00+00:00",

"canonical_url": "https://x.com/aleabitoreddit/status/999999999999999999",

"needle": "canonical_url 指向非 X 原帖来源",                     "row": {                         "canonical_url": "https://www.sotwe.com/ALEABITOREDDIT",

"row": {                         "canonical_url": "https://www.sotwe.com/ALEABITOREDDIT",                         "posted_at": "unknown",

"canonical_url": "https://www.sotwe.com/ALEABITOREDDIT",

"needle": "source_status 不合格",                     "row": {                         "canonical_url": "https://x.com/aleabitoreddit/status/999999999999999998",

"row": {                         "canonical_url": "https://x.com/aleabitoreddit/status/999999999999999998",                         "posted_at": "2026-06-08T05:10:00+00:00",

"canonical_url": "https://x.com/aleabitoreddit/status/999999999999999998",

"needle": "source_status 不合格",                     "row": {                         "canonical_url": "https://x.com/aleabitoreddit/status/999999999999999997",

"row": {                         "canonical_url": "https://x.com/aleabitoreddit/status/999999999999999997",                         "posted_at": "2026-06-08T05:20:00+00:00",

"canonical_url": "https://x.com/aleabitoreddit/status/999999999999999997",

"youtube",                     "--locator",                     "https://www.youtube.com/@channel/videos",

"--locator",                     "https://www.youtube.com/@channel/videos",                 ]

"https://www.youtube.com/@channel/videos",

"x",                     "--locator",                     "https://x.com/x_only_demo",

"--locator",                     "https://x.com/x_only_demo",                 ]

"https://x.com/x_only_demo",

"website",                     "--locator",                     "https://example.com",

"--locator",                     "https://example.com",                 ]

"https://example.com",

帮我安装这个 skill：https://github.com/AlphaMao1/AlphaMao_Skills/tree/main/skills/intuition-pumps

<http://xhslink.com/o/5LGsU9I2FoV>

帮我安装这个 skill：https://github.com/AlphaMao1/AlphaMao_Skills/tree/main/skills/market-sizing

<http://xhslink.com/o/9zGJ009rm21>

"unit": "people",             "value_or_path": "1.408B",             "url_or_endpoint": "https://api.worldbank.org/v2/country/CN/indicator/SP.POP.TOTL",

"value_or_path": "1.408B",             "url_or_endpoint": "https://api.worldbank.org/v2/country/CN/indicator/SP.POP.TOTL",             "accessed_at": "2026-06-24",

"url_or_endpoint": "https://api.worldbank.org/v2/country/CN/indicator/SP.POP.TOTL",

Executable file

Executable file

"unit": "people",                 "value_or_path": "API series",                 "url_or_endpoint": "https://api.worldbank.org/v2/country/CN/indicator/SP.POP.TOTL",

"value_or_path": "API series",                 "url_or_endpoint": "https://api.worldbank.org/v2/country/CN/indicator/SP.POP.TOTL",                 "accessed_at": "2026-06-24",

"url_or_endpoint": "https://api.worldbank.org/v2/country/CN/indicator/SP.POP.TOTL",

- World Bank API: `https://api.worldbank.org/v2/`

- FRED API: `https://fred.stlouisfed.org/docs/api/fred/`

- DBnomics: `https://db.nomics.world/`

- OECD Data API: `https://www.oecd.org/en/data/insights/data-explainers/2024/09/api.html`

- IMF DataMapper API: `https://www.imf.org/external/datamapper/api/help`

- UN Comtrade API: `https://comtrade.un.org/`

- UN Comtrade Python package: `https://github.com/uncomtrade/comtradeapicall`

- SEC EDGAR APIs: `https://www.sec.gov/edgar/sec-api-documentation`

Executable file

"provider": "World Bank",         "best_for": "population, GDP, macro indicators across countries",         "endpoint": "https://api.worldbank.org/v2/",

"best_for": "population, GDP, macro indicators across countries",         "endpoint": "https://api.worldbank.org/v2/",         "notes": "No API key. Good baseline for country-level denominators.",

"endpoint": "https://api.worldbank.org/v2/",

"provider": "Federal Reserve Economic Data",         "best_for": "US macro and financial time series",         "endpoint": "https://fred.stlouisfed.org/docs/api/fred/",

"best_for": "US macro and financial time series",         "endpoint": "https://fred.stlouisfed.org/docs/api/fred/",         "notes": "Requires a free API key for package use.",

"endpoint": "https://fred.stlouisfed.org/docs/api/fred/",

"provider": "DBnomics",         "best_for": "aggregated public macro datasets from many providers",         "endpoint": "https://api.db.nomics.world/v22/",

"best_for": "aggregated public macro datasets from many providers",         "endpoint": "https://api.db.nomics.world/v22/",         "notes": "Useful discovery layer when provider-specific APIs are fragmented.",

"endpoint": "https://api.db.nomics.world/v22/",

"provider": "OECD",         "best_for": "OECD structural, industry, education, labor, and policy indicators",         "endpoint": "https://sdmx.oecd.org/public/rest/",

"best_for": "OECD structural, industry, education, labor, and policy indicators",         "endpoint": "https://sdmx.oecd.org/public/rest/",         "notes": "Use SDMX queries; record dataset and dimension filters in source cards.",

"endpoint": "https://sdmx.oecd.org/public/rest/",

"provider": "IMF",         "best_for": "WEO, macro, finance, country and region comparisons",         "endpoint": "https://www.imf.org/external/datamapper/api/",

"best_for": "WEO, macro, finance, country and region comparisons",         "endpoint": "https://www.imf.org/external/datamapper/api/",         "notes": "DataMapper is convenient for indicators and country sets.",

"endpoint": "https://www.imf.org/external/datamapper/api/",

"provider": "UN Comtrade",         "best_for": "trade, HS codes, import/export side checks",         "endpoint": "https://comtradeapi.un.org/",

"best_for": "trade, HS codes, import/export side checks",         "endpoint": "https://comtradeapi.un.org/",         "notes": "Good for hardware, commodity, equipment, and cross-border checks.",

"endpoint": "https://comtradeapi.un.org/",

"provider": "National Bureau of Statistics of China",         "best_for": "China population, macro, industry output, retail, fixed asset data",         "endpoint": "https://data.stats.gov.cn/",

"best_for": "China population, macro, industry output, retail, fixed asset data",         "endpoint": "https://data.stats.gov.cn/",         "notes": "Prefer original tables; AkShare can be a convenience wrapper.",

"endpoint": "https://data.stats.gov.cn/",

"provider": "SEC EDGAR / CNInfo / exchange disclosures",         "best_for": "competitor revenue, segment revenue, customer and geography exposure",         "endpoint": "https://www.sec.gov/edgar/sec-api-documentation",

"best_for": "competitor revenue, segment revenue, customer and geography exposure",         "endpoint": "https://www.sec.gov/edgar/sec-api-documentation",         "notes": "Use filings as primary evidence when market claims depend on listed

"endpoint": "https://www.sec.gov/edgar/sec-api-documentation",

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

params["date"] = f"{start_year}:{end_year}"         url = (             f"https://api.worldbank.org/v2/country/{country}/indicator/{indicator}"

url = (             f"https://api.worldbank.org/v2/country/{country}/indicator/{indicator}"             f"?{urlencode(params)}"

f"https://api.worldbank.org/v2/country/{country}/indicator/{indicator}"

def get_dbnomics_series(self, provider: str, dataset: str, series: str) -> pd.DataFrame:         url = f"https://api.db.nomics.world/v22/series/{provider}/{dataset}/{series}"

url = f"https://api.db.nomics.world/v22/series/{provider}/{dataset}/{series}"

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

Executable file

Executable file

Executable file

[Credential-related source omitted]

帮我安装这个 skill： https://github.com/AlphaMao1/AlphaMao_Skills/tree/main/skills/notion-course-pack-init

https://github.com/AlphaMao1/AlphaMao_Skills/tree/main/skills/notion-course-pack-init

git clone https://github.com/AlphaMao1/AlphaMao_Skills.git

https://mcp.notion.com/mcp

- [Notion MCP](https://developers.notion.com/guides/mcp/overview)

- [Connecting to Notion MCP](https://developers.notion.com/guides/mcp/get-started-with-mcp)

- [Apps in ChatGPT](https://help.openai.com/en/articles/11487775-connectors-in-chatgpt)

- [Notion for education](https://www.notion.com/help/notion-for-education): education workspace setup, templates, student/teacher workspace assumptions.

- [Setting up Notion for school](https://www.notion.com/help/guides/setting-up-notion-for-school): school workspace, notes, reading lists, planning, collaboration.

- [Start with a template](https://www.notion.com/help/guides/start-with-a-template): templates as starting structure, not as content to copy blindly.

- [Page icons & covers](https://www.notion.com/help/guides/page-icons-and-covers): icons and covers help page recognition and digestibility.

- [Using database views](https://www.notion.com/help/guides/using-database-views): table, list, board, gallery, calendar, and timeline views.

- [When to use each type of database view](https://www.notion.com/help/guides/when-to-use-each-type-of-database-view): choose views based on task, not decoration.

- [Board view databases](https://www.notion.com/help/guides/board-view-databases): boards work for status-based task and process management.

- [Getting started with projects and tasks](https://www.notion.com/help/guides/getting-started-with-projects-and-tasks): simple task systems, projects/tasks relation, avoid over-heavy process.

- [Give your to-dos a home with task databases](https://www.notion.com/help/guides/give-your-to-dos-a-home-with-task-databases): central task surfaces and filtered views.

- [Dashboards view](https://www.notion.com/help/dashboards): useful as an information-architecture reference, but do not require it because it is Business/Enterprise plan functionality.

- [Student dashboards category](https://www.notion.com/templates/category/student-dashboards): home base for courses, assignments, exams, deadlines, progress.

- [Free course site templates category](https://www.notion.com/templates/category/free-course-site-templates): course sites, class hubs, syllabi, course delivery pages.

- [Course Delivery Hub / Course Portal](https://www.notion.com/templates/course-delivery-hub-course-portal): course overview plus module/course delivery navigation.

- [Knowledge Hub Template](https://www.notion.com/templates/simple-knowledge-hub): notes, references, and insights in one hub.

- [Top free Knowledge Hub templates](https://www.notion.com/templates/collections/top-10-free-knowledge-hub-templates-in-notion): searchability, taxonomy, and avoiding over-complex structures.

- [Minimalist Student Dashboard](https://www.notion.com/templates/minimalist-student-dashboard): simple course, assignment, exam, and note arrangement.

- [Janice Studies' student dashboard](https://www.notion.com/templates/student-dashboard): tasks, to-dos, courses, notes, and assignment awareness in one place.

[mcp_servers.notion] url = "https://mcp.notion.com/mcp"

url = "https://mcp.notion.com/mcp"

- Notion MCP overview: https://developers.notion.com/guides/mcp/overview

- Notion MCP connection guide: https://developers.notion.com/guides/mcp/get-started-with-mcp

- OpenAI apps guidance: https://help.openai.com/en/articles/11487775-connectors-in-chatgpt

- Notion's hosted MCP server supports user-authorized read/write access and documents Codex configuration through `https://mcp.notion.com/mcp`: https://developers.notion.com/guides/mcp/get-started-with-mcp

- OpenAI calls integrations "apps": https://help.openai.com/en/articles/11487775-connectors-in-chatgpt

课程首页 URL：https://app.notion.com/p/00000000000000000000000000000009

- 课程名称：Fixture Course - 课程首页：https://app.notion.com/p/00000000000000000000000000000001

- 课程首页：https://app.notion.com/p/00000000000000000000000000000001

},   "notion": {     "parent_url": "https://app.notion.com/p/00000000000000000000000000000000",

"notion": {     "parent_url": "https://app.notion.com/p/00000000000000000000000000000000",     "course_home_url": "https://app.notion.com/p/00000000000000000000000000000001",

"parent_url": "https://app.notion.com/p/00000000000000000000000000000000",

"course_home_url": "https://app.notion.com/p/00000000000000000000000000000001",

"pages": {       "course_home": "https://app.notion.com/p/00000000000000000000000000000001",       "runtime_snapshot": "https://app.notion.com/p/00000000000000000000000000000002",

"course_home": "https://app.notion.com/p/00000000000000000000000000000001",

"runtime_snapshot": "https://app.notion.com/p/00000000000000000000000000000002",

"course_map": "https://app.notion.com/p/00000000000000000000000000000003",

"course_state_profile": "https://app.notion.com/p/00000000000000000000000000000004",

"sources_coverage": "https://app.notion.com/p/00000000000000000000000000000005",

"sessions": "https://app.notion.com/p/00000000000000000000000000000006",

"notes_inbox": "https://app.notion.com/p/00000000000000000000000000000007",

"codex_handoff_queue": "https://app.notion.com/p/00000000000000000000000000000008"

},     "parent_child": {       "course_home_parent_url": "https://app.notion.com/p/00000000000000000000000000000000",

"parent_child": {       "course_home_parent_url": "https://app.notion.com/p/00000000000000000000000000000000",       "children_of_course_home": [

"course_home_parent_url": "https://app.notion.com/p/00000000000000000000000000000000",

"course_home": {           "page_role": "course_home",           "url": "https://app.notion.com/p/00000000000000000000000000000001",

"page_role": "course_home",           "url": "https://app.notion.com/p/00000000000000000000000000000001",           "title": "Fixture Course",

"url": "https://app.notion.com/p/00000000000000000000000000000001",

"runtime_snapshot": {           "page_role": "runtime_snapshot",           "url": "https://app.notion.com/p/00000000000000000000000000000002",

"page_role": "runtime_snapshot",           "url": "https://app.notion.com/p/00000000000000000000000000000002",           "title": "Runtime Snapshot",

"url": "https://app.notion.com/p/00000000000000000000000000000002",

"course_map": {           "page_role": "course_map",           "url": "https://app.notion.com/p/00000000000000000000000000000003",

"page_role": "course_map",           "url": "https://app.notion.com/p/00000000000000000000000000000003",           "title": "Course_Map",

"url": "https://app.notion.com/p/00000000000000000000000000000003",

"course_state_profile": {           "page_role": "course_state_profile",           "url": "https://app.notion.com/p/00000000000000000000000000000004",

"page_role": "course_state_profile",           "url": "https://app.notion.com/p/00000000000000000000000000000004",           "title": "Course_State & Profile",

"url": "https://app.notion.com/p/00000000000000000000000000000004",

"sources_coverage": {           "page_role": "sources_coverage",           "url": "https://app.notion.com/p/00000000000000000000000000000005",

"page_role": "sources_coverage",           "url": "https://app.notion.com/p/00000000000000000000000000000005",           "title": "Sources & Coverage",

"url": "https://app.notion.com/p/00000000000000000000000000000005",

"sessions": {           "page_role": "sessions",           "url": "https://app.notion.com/p/00000000000000000000000000000006",

"page_role": "sessions",           "url": "https://app.notion.com/p/00000000000000000000000000000006",           "title": "Sessions",

"url": "https://app.notion.com/p/00000000000000000000000000000006",

"notes_inbox": {           "page_role": "notes_inbox",           "url": "https://app.notion.com/p/00000000000000000000000000000007",

"page_role": "notes_inbox",           "url": "https://app.notion.com/p/00000000000000000000000000000007",           "title": "Notes Inbox",

"url": "https://app.notion.com/p/00000000000000000000000000000007",

"codex_handoff_queue": {           "page_role": "codex_handoff_queue",           "url": "https://app.notion.com/p/00000000000000000000000000000008",

"page_role": "codex_handoff_queue",           "url": "https://app.notion.com/p/00000000000000000000000000000008",           "title": "Codex Handoff Queue",

"url": "https://app.notion.com/p/00000000000000000000000000000008",

"test_course_id": "system_check_fixture_course",     "test_course_title": "SYSTEM CHECK | Fixture Course Runtime",     "test_course_home_url": "https://app.notion.com/p/00000000000000000000000000000009",

"test_course_title": "SYSTEM CHECK | Fixture Course Runtime",     "test_course_home_url": "https://app.notion.com/p/00000000000000000000000000000009",     "isolated_parent_url": "https://app.notion.com/p/00000000000000000000000000000010",

"test_course_home_url": "https://app.notion.com/p/00000000000000000000000000000009",

"isolated_parent_url": "https://app.notion.com/p/00000000000000000000000000000010",

"runtime_snapshot": {         "page_role": "runtime_snapshot",         "url": "https://app.notion.com/p/00000000000000000000000000000011",

"page_role": "runtime_snapshot",         "url": "https://app.notion.com/p/00000000000000000000000000000011",         "managed_region_test_id_occurrences": 1,

"url": "https://app.notion.com/p/00000000000000000000000000000011",

"course_state_profile": {         "page_role": "course_state_profile",         "url": "https://app.notion.com/p/00000000000000000000000000000012",

"page_role": "course_state_profile",         "url": "https://app.notion.com/p/00000000000000000000000000000012",         "managed_region_test_id_occurrences": 1,

"url": "https://app.notion.com/p/00000000000000000000000000000012",

"sessions": {         "page_role": "sessions",         "url": "https://app.notion.com/p/00000000000000000000000000000013",

"page_role": "sessions",         "url": "https://app.notion.com/p/00000000000000000000000000000013",         "managed_region_test_id_occurrences": 1,

"url": "https://app.notion.com/p/00000000000000000000000000000013",

"notes_inbox": {         "page_role": "notes_inbox",         "url": "https://app.notion.com/p/00000000000000000000000000000014",

"page_role": "notes_inbox",         "url": "https://app.notion.com/p/00000000000000000000000000000014",         "managed_region_test_id_occurrences": 1,

"url": "https://app.notion.com/p/00000000000000000000000000000014",

"course_map": {         "page_role": "course_map",         "url": "https://app.notion.com/p/00000000000000000000000000000015",

"page_role": "course_map",         "url": "https://app.notion.com/p/00000000000000000000000000000015",         "before_fingerprint": "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa",

"url": "https://app.notion.com/p/00000000000000000000000000000015",

"sources_coverage": {         "page_role": "sources_coverage",         "url": "https://app.notion.com/p/00000000000000000000000000000016",

"page_role": "sources_coverage",         "url": "https://app.notion.com/p/00000000000000000000000000000016",         "before_fingerprint": "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb",

"url": "https://app.notion.com/p/00000000000000000000000000000016",

课程首页 URL：https://app.notion.com/p/00000000000000000000000000000009

#!/usr/bin/env python3

Executable file

"test_course_id": "system-check-1",             "test_course_title": "SYSTEM CHECK | 课程一",             "test_course_home_url": "https://app.notion.com/system-check/home",

"test_course_title": "SYSTEM CHECK | 课程一",             "test_course_home_url": "https://app.notion.com/system-check/home",             "isolated_parent_url": "https://app.notion.com/system-check/parent",

"test_course_home_url": "https://app.notion.com/system-check/home",

"isolated_parent_url": "https://app.notion.com/system-check/parent",

key: {                     "page_role": key,                     "url": f"https://app.notion.com/system-check/{key}",

"page_role": key,                     "url": f"https://app.notion.com/system-check/{key}",                     "managed_region_test_id_occurrences": 1,

"url": f"https://app.notion.com/system-check/{key}",

key: {                     "page_role": key,                     "url": f"https://app.notion.com/system-check/{key}",

"page_role": key,                     "url": f"https://app.notion.com/system-check/{key}",                     "before_fingerprint": sha(f"protected-{key}"),

"url": f"https://app.notion.com/system-check/{key}",

make_course_design(course_design_path, course_id="course-1")     pages = {         key: f"https://app.notion.com/course/{index}"

pages = {         key: f"https://app.notion.com/course/{index}"         for index, key in enumerate(RENDERER.REQUIRED_PAGE_KEYS, start=1)

key: f"https://app.notion.com/course/{index}"

course_title="课程一",         created_at="2026-07-11T00:00:00+00:00",         parent_url="https://app.notion.com/parent",

created_at="2026-07-11T00:00:00+00:00",         parent_url="https://app.notion.com/parent",         template_version="v3",

parent_url="https://app.notion.com/parent",

with tempfile.TemporaryDirectory() as temp:             args, receipt_path, receipt = make_args(Path(temp))             fake_pages = {key: f"https://fake.invalid/{key}" for key in RENDERER.REQUIRED_PAGE_KEYS}

args, receipt_path, receipt = make_args(Path(temp))             fake_pages = {key: f"https://fake.invalid/{key}" for key in RENDERER.REQUIRED_PAGE_KEYS}             zero_fingerprints = {key: "0" * 64 for key in RENDERER.REQUIRED_PAGE_KEYS}

fake_pages = {key: f"https://fake.invalid/{key}" for key in RENDERER.REQUIRED_PAGE_KEYS}

Executable file

target, manifest_path, manifest = self.copy_fixture(temp)             page = "course_home"             manifest["notion"]["pages"][page] = "https://fake.invalid/course-home"

page = "course_home"             manifest["notion"]["pages"][page] = "https://fake.invalid/course-home"             manifest["notion"]["course_home_url"] = "https://fake.invalid/course-home"

manifest["notion"]["pages"][page] = "https://fake.invalid/course-home"

manifest["notion"]["course_home_url"] = "https://fake.invalid/course-home"

manifest["notion"]["managed_regions"][page]["source_fingerprint"] = "0" * 64             manifest["notion"]["readback"]["pages"][page]["url"] = (                 "https://fake.invalid/course-home"

manifest["notion"]["readback"]["pages"][page]["url"] = (                 "https://fake.invalid/course-home"             )

"https://fake.invalid/course-home"

smoke.setdefault("write_page_receipts", {})["sessions"] = {                 "page_role": "sessions",                 "url": "https://app.notion.com/system-check/sessions",

"page_role": "sessions",                 "url": "https://app.notion.com/system-check/sessions",                 "managed_region_test_id_occurrences": 0,

"url": "https://app.notion.com/system-check/sessions",

smoke.setdefault("protected_page_receipts", {})["course_map"] = {                 "page_role": "course_map",                 "url": "https://app.notion.com/system-check/course-map",

"page_role": "course_map",                 "url": "https://app.notion.com/system-check/course-map",                 "before_fingerprint": "a" * 64,

"url": "https://app.notion.com/system-check/course-map",

Executable file

#!/usr/bin/env python3

Executable file

setup = setup_path.read_text(encoding="utf-8")         for required in (             "https://mcp.notion.com/mcp",

for required in (             "https://mcp.notion.com/mcp",             "codex mcp login notion",

"https://mcp.notion.com/mcp",

帮我安装这个 skill：https://github.com/AlphaMao1/AlphaMao_Skills/tree/main/skills/pick-movie-theater

影片端分开记录 `Filmed for IMAX`、IMAX DMR、1.90:1／1.43:1 扩展画幅、扩画时长、专用声音和 IMAX 70mm 胶片拷贝。IMAX 官方说明 `Filmed for IMAX` 会在创作到后期阶段合作，并可能采用 1.90:1 或 1.43:1；这不表示每部相关影片都有 1.43:1 或胶片版：[IMAX 官方说明](https://www.imax.com/filmed-for-imax)。

| 购票标签 | 实际保证或线索 | 不自动保证 | | --- | --- | --- | | Dolby Cinema／杜比影院 | 完整品牌方案；核实官方认证。Dolby 将其描述为 Dolby Vision 双 4K 激光、Dolby Atmos 和整体影厅设计的组合：[Dolby Cinema](https://www.dolby.com/experience/cinema/) | 不保证本片一定有 Atmos 原生混音，也不保证银幕比例和本片画幅天然匹配 |

| --- | --- | --- | | Dolby Cinema／杜比影院 | 完整品牌方案；核实官方认证。Dolby 将其描述为 Dolby Vision 双 4K 激光、Dolby Atmos 和整体影厅设计的组合：[Dolby Cinema](https://www.dolby.com/experience/cinema/) | 不保证本片一定有 Atmos 原生混音，也不保证银幕比例和本片画幅天然匹配 | | Dolby Vision + Dolby Atmos 

| Dolby Cinema／杜比影院 | 完整品牌方案；核实官方认证。Dolby 将其描述为 Dolby Vision 双 4K 激光、Dolby Atmos 和整体影厅设计的组合：[Dolby Cinema](https://www.dolby.com/experience/cinema/) | 不保证本片一定有 Atmos 原生混音，也不保证银幕比例和本片画幅天然匹配 |

| Dolby Vision + Dolby Atmos 高级厅 | 同时出现两项能力时继续核实授权、投影和厅级方案 | 不因两个词同时出现就升级为 Dolby Cinema | | Dolby Atmos 厅／杜比全景声厅 | 对象式沉浸音频和头顶扬声器能力：[Dolby Atmos 影院说明](https://professional.dolby.com/cinema/dolby-atmos/) | 不保证 Dolby Vision、双 4K、HDR、巨幕或本片有 Atm

| Dolby Atmos 厅／杜比全景声厅 | 对象式沉浸音频和头顶扬声器能力：[Dolby Atmos 影院说明](https://professional.dolby.com/cinema/dolby-atmos/) | 不保证 Dolby Vision、双 4K、HDR、巨幕或本片有 Atmos 音轨 |

- CINITY：分别核实影片是否有专用 4K、HFR、HDR、WCG、高亮或 3D 版本，以及具体厅当前系统能力。系统支持最高 4K／120fps 不表示每部影片都获得该版本：[CINITY 官方技术页](https://cinity.com/)。

THX 曾给影院设计者以最远座约 36° 水平视角作为认证银幕布置参考；它是设计参考，不是所有影片、银幕和用户的唯一最佳座位：[THX screen placement](https://www.thx.com/questions/thx-certified-screen-placement/)。不要把某个历史角度或“全厅三分之二排”写成普适硬标准。

优先影厅横向中轴。Dolby Atmos 影院规范以参考聆听位（RLP）为中心定义中央聆听区域（CLA），典型 CLA 宽度为影厅宽度的三分之一、长度为首末排距离的三分之一：[Dolby Atmos Specifications](https://professional.dolby.com/siteassets/cinema/dolby-audio-products/dolby-atmos-specifications.pdf)。

#!/usr/bin/env python3

- 中国日报，CINITY LED 独占 4K 3D 48fps：https://cn.chinadaily.com.cn/a/202511/27/WS69280fb1a310942cc4993bfb.html

- 东来印象 CINITY LED 厅参数与开业排片：https://finance.sina.cn/2025-12-08/detail-inhaavvf1696662.d.html

- 成都两家 CINITY LED 首轮兼容故障：https://news.sina.cn/2025-12-20/detail-inhcmwzi1215141.d.html?vt=4

- 环球中心双制式升级资料：https://finance.sina.cn/tech/2022-11-03/detail-imqqsmrp4747650.d.html

- 环球中心《阿凡达：火与烬》CINITY 实际观影报道：https://k.sina.com.cn/article_1959451603_74cadbd302701gy4q.html?from=movie

- Dolby 官方院线片单：https://professional.dolby.com/zh-cn/cinema/theatrical-releases/

- IMAX 官方影片页：https://www.imax.com/movie/avatar-fire-and-ash

- 中国内地版本与画幅分析：https://m.thepaper.cn/newsDetail_forward_32201479

- IMAX 影片页：https://www.imax.com/theodyssey

- MCL 影片页：https://services.mclcinema.com/Ticketing/MovieSet?MovieSetID=14750&language=en-US

- MCL 影院资料：https://www.mclcinema.com.hk/mclaboutus.aspx?vislang=2

- 当日全港排片快照：https://m.wmoov.com/movie/showtimes/70142?version=IMAX

- https://chinafilmnews.cn/Html/2026-07-01/21279.html

- https://www.ithome.com/0/981/540.htm

1. **影片端方向**：IMAX 官方确认影片全片使用 IMAX 胶片摄影机拍摄；不能据此推断上海任何 IMAX 都有 1.43 或 IMAX 70mm。 2. **待核验候选**：正大乐影城公开报道显示其 IMAX 银幕为 24.38 米 × 13.6 米，配新一代激光与 12.1 声道；这证明影厅能力线索，不证明《奥德赛》当地版本和具体排片。候选来源：https://www.esquirecn.cn/xiansheng/detail/49850

2. **待核验候选**：正大乐影城公开报道显示其 IMAX 银幕为 24.38 米 × 13.6 米，配新一代激光与 12.1 声道；这证明影厅能力线索，不证明《奥德赛》当地版本和具体排片。候选来源：https://www.esquirecn.cn/xiansheng/detail/49850

Executable file

[Credential-related source omitted]

[小红书视频：我把 Jev 接进了持仓监控 Skill](https://xhslink.cn/o/1TlpO5emmFE)

帮我安装这个 Skill：https://github.com/AlphaMao1/AlphaMao_Skills/tree/main/skills/position-thesis-monitor

[Credential-related source omitted]

[Credential-related source omitted]

"adjustment": "raw",       "provider": "synthetic-example",       "source_url": "https://example.com/demo-market",

"provider": "synthetic-example",       "source_url": "https://example.com/demo-market",       "values": {

"source_url": "https://example.com/demo-market",

"family": "issuer:demo",       "kind": "primary",       "url_prefix": "https://example.com/disclosures/",

"kind": "primary",       "url_prefix": "https://example.com/disclosures/",       "cadence": "manual"

"url_prefix": "https://example.com/disclosures/",

"baseline": [],   "sources": [     {"id": "sec", "family": "issuer:1045810", "kind": "primary", "url_prefix": "https://www.sec.gov/Archives/edgar/data/1045810/", "cadence": "daily"}

"sources": [     {"id": "sec", "family": "issuer:1045810", "kind": "primary", "url_prefix": "https://www.sec.gov/Archives/edgar/data/1045810/", "cadence": "daily"}   ],

{"id": "sec", "family": "issuer:1045810", "kind": "primary", "url_prefix": "https://www.sec.gov/Archives/edgar/data/1045810/", "cadence": "daily"}

官方入口：[API](https://docs.typesafe.ai/api.md)、[模型限制](https://docs.typesafe.ai/models.md)、[并行问题](https://docs.typesafe.ai/cookbooks/parallel_questions.md)、[候选值抽取](https://docs.typesafe.ai/cookbooks/pre_parsed_value_extraction_cookbook.md)、[已知局

{   "source_id": "sec",   "source_url": "https://www.sec.gov/Archives/edgar/data/1045810/...",

"source_id": "sec",   "source_url": "https://www.sec.gov/Archives/edgar/data/1045810/...",   "origin_id": "sec:真实accession",

"source_url": "https://www.sec.gov/Archives/edgar/data/1045810/...",

[Credential-related source omitted]

Executable file

require(available, "Filing needs an acceptance timestamp")             require(stamp(available) <= stamp(as_of), "Future filing rejected")             receipt = fetch(filing["url"], ["www.sec.gov"], engine.workspace / "captures")

require(stamp(available) <= stamp(as_of), "Future filing rejected")             receipt = fetch(filing["url"], ["www.sec.gov"], engine.workspace / "captures")             text = (engine.workspace / "captures" / receipt["text_file"]).read_te

receipt = fetch(filing["url"], ["www.sec.gov"], engine.workspace / "captures")

Executable file

Executable file

#!/usr/bin/env python3

#!/usr/bin/env python3

data = json.dumps(payload, ensure_ascii=False).encode("utf-8")                 headers["Content-Type"] = "application/json"             request = urllib.request.Request(url, data=data, headers=headers)

headers["Content-Type"] = "application/json"             request = urllib.request.Request(url, data=data, headers=headers)             with urllib.request.urlopen(request, timeout=15) as response:

request = urllib.request.Request(url, data=data, headers=headers)

with urllib.request.urlopen(request, timeout=15) as response:

security_kind: str, ) -> dict[str, Any]:     source_url = urllib.parse.urljoin("https://www.sse.com.cn", row["URL"])

) -> dict[str, Any]:     source_url = urllib.parse.urljoin("https://www.sse.com.cn", row["URL"])     published = row.get("ADDDATE") or f"{row['SSEDATE']} 00:00:00"

source_url = urllib.parse.urljoin("https://www.sse.com.cn", row["URL"])

"endDate": end.isoformat(),     }     endpoint = "https://query.sse.com.cn/security/stock/queryCompanyBulletin.do"

}     endpoint = "https://query.sse.com.cn/security/stock/queryCompanyBulletin.do"     rows, query_url = _sse_pages(

endpoint = "https://query.sse.com.cn/security/stock/queryCompanyBulletin.do"

endpoint,         params,         referer="https://www.sse.com.cn/disclosure/listedinfo/announcement/",

params,         referer="https://www.sse.com.cn/disclosure/listedinfo/announcement/",     )

referer="https://www.sse.com.cn/disclosure/listedinfo/announcement/",

"CODE_ASC": "",     }     endpoint = "https://query.sse.com.cn/commonQuery.do"

}     endpoint = "https://query.sse.com.cn/commonQuery.do"     rows, query_url = _sse_pages(

endpoint = "https://query.sse.com.cn/commonQuery.do"

endpoint,         params,         referer="https://www.sse.com.cn/disclosure/fund/announcement/",

params,         referer="https://www.sse.com.cn/disclosure/fund/announcement/",     )

referer="https://www.sse.com.cn/disclosure/fund/announcement/",

security_kind: str, ) -> dict[str, Any]:     source_url = urllib.parse.urljoin("https://disc.static.szse.cn", row["attachPath"])

) -> dict[str, Any]:     source_url = urllib.parse.urljoin("https://disc.static.szse.cn", row["attachPath"])     source_id_raw = row.get("annId")

source_url = urllib.parse.urljoin("https://disc.static.szse.cn", row["attachPath"])

security_kind: str, ) -> tuple[list[dict[str, Any]], str, str]:     url = "https://www.szse.cn/api/disc/announcement/annList"

) -> tuple[list[dict[str, Any]], str, str]:     url = "https://www.szse.cn/api/disc/announcement/annList"     channel = "fundinfoNotice_disc" if security_kind == "fund" else "listedNotice_disc"

url = "https://www.szse.cn/api/disc/announcement/annList"

channel = "fundinfoNotice_disc" if security_kind == "fund" else "listedNotice_disc"     referer = (         "https://www.szse.cn/disclosure/fund/notice/index.html"

referer = (         "https://www.szse.cn/disclosure/fund/notice/index.html"         if security_kind == "fund"

"https://www.szse.cn/disclosure/fund/notice/index.html"

if security_kind == "fund"         else "https://www.szse.cn/disclosure/listed/notice/index.html"     )

else "https://www.szse.cn/disclosure/listed/notice/index.html"

#!/usr/bin/env python3

for attempt in range(attempts):         try:             request = urllib.request.Request(url, headers={"User-Agent": UA, "Referer": url})

try:             request = urllib.request.Request(url, headers={"User-Agent": UA, "Referer": url})             with urllib.request.urlopen(request, timeout=timeout) as response:

request = urllib.request.Request(url, headers={"User-Agent": UA, "Referer": url})

with urllib.request.urlopen(request, timeout=timeout) as response:

def fetch_tencent(info: dict[str, str], history: int) -> dict[str, Any]:     url = "https://web.ifzq.gtimg.cn/appstock/app/fqkline/get?" + urllib.parse.urlencode(

url = "https://web.ifzq.gtimg.cn/appstock/app/fqkline/get?" + urllib.parse.urlencode(

for host in ("push2his.eastmoney.com", "push2graydelay.eastmoney.com"):         candidate = f"https://{host}/api/qt/stock/kline/get?" + urllib.parse.urlencode(params)

candidate = f"https://{host}/api/qt/stock/kline/get?" + urllib.parse.urlencode(params)

#!/usr/bin/env python3

SUBMISSIONS_ROOT = "https://data.sec.gov/submissions"

TICKER_MAP_URL = "https://www.sec.gov/files/company_tickers.json"

ARCHIVES_ROOT = "https://www.sec.gov/Archives/edgar/data"

for attempt in range(3):         try:             request = urllib.request.Request(

try:             request = urllib.request.Request(                 url,

request = urllib.request.Request(

},             )             with urllib.request.urlopen(request, timeout=20) as response:

)             with urllib.request.urlopen(request, timeout=20) as response:                 body = response.read()

with urllib.request.urlopen(request, timeout=20) as response:

#!/usr/bin/env python3

Executable file

API_URL = "https://api.typesafe.ai/v1/systemone"

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

Executable file

Executable file

Executable file

from urllib.parse import urlparse             if not args.url.startswith(source["url_prefix"]): raise ValueError("Capture is outside registered source prefix")             result = fetch(args.url, [urlparse(source["url_prefix"]).hostname], 

if not args.url.startswith(source["url_prefix"]): raise ValueError("Capture is outside registered source prefix")             result = fetch(args.url, [urlparse(source["url_prefix"]).hostname], args.workspace / "captures")         elif args

result = fetch(args.url, [urlparse(source["url_prefix"]).hostname], args.workspace / "captures")

Executable file

Executable file

def fetch(url, allowed_hosts, output_dir):

[Credential-related source omitted]

Executable file

Executable file

def sample_plan():     return {"id": "CASE", "mode": "simulation", "effective_at": "2025-01-01T00:00:00+00:00", "hypotheses": [{"id": "H1", "claim": "Data center demand is growing", "counter_condition": "A realized decrease in data center d

return {"id": "CASE", "mode": "simulation", "effective_at": "2025-01-01T00:00:00+00:00", "hypotheses": [{"id": "H1", "claim": "Data center demand is growing", "counter_condition": "A realized decrease in data center demand", "status": "draf

def event(text="The company reports unchanged operating conditions. No special event occurred.", origin="filing-1", date=ASOF, period="FY2025"):     return {"source_id": "sec", "origin_id": origin, "source_url": "https://www.sec.gov/Archive

return {"source_id": "sec", "origin_id": origin, "source_url": "https://www.sec.gov/Archives/edgar/data/1045810/test.htm", "available_at": date, "text": text, "period": period}

def test_future_baseline_rejected(self):         self.plan["baseline"] = [{"summary": "future", "source_url": "https://example.com", "available_at": ASOF}]

self.plan["baseline"] = [{"summary": "future", "source_url": "https://example.com", "available_at": ASOF}]

def test_out_of_scope_source_rejected(self):         data = event(); data["source_url"] = "https://evil.example/"

data = event(); data["source_url"] = "https://evil.example/"

Executable file

values={'close':{'value':price,'unit':'USD'},'daily_return':{'value':ret,'unit':'ratio'},'volume':{'value':1000,'unit':'shares'}}     if benchmark is not None:values['benchmark_return']={'value':benchmark,'unit':'ratio'}     return {'as_of'

return {'as_of':date,'currency':'USD','adjustment':'raw','provider':'fixture','source_url':'https://example.com/prices','values':values}

[Credential-related source omitted]

帮我安装这个 skill：https://github.com/AlphaMao1/AlphaMao_Skills/tree/main/skills/progressive-investment-research

<http://xhslink.com/o/4GUbrB6dvLr>

| 通用发现 | AnySearch / web search / Tavily / runtime search | 找来源入口、官方页面、新闻线索、大数 | 不再做新通用搜索器 | | 登录态和动态页面 | web-access / browser CDP / user-authorized browser | 用户授权登录态、动态渲染、交互页面 | 不再做通用浏览器抓取层 | | 已知 URL 正文 | AnySearch `extract` / Jina / WebF

| 登录态和动态页面 | web-access / browser CDP / user-authorized browser | 用户授权登录态、动态渲染、交互页面 | 不再做通用浏览器抓取层 | | 已知 URL 正文 | AnySearch `extract` / Jina / WebFetch / curl / runtime URL fetch | HTML 正文化、网页正文 | 不把正文抽取当事实核验 | | 本地文件读取 | markitdown / pdf /

| 已知 URL 正文 | AnySearch `extract` / Jina / WebFetch / curl / runtime URL fetch | HTML 正文化、网页正文 | 不把正文抽取当事实核验 |

python <anysearch>/scripts/anysearch_cli.py search "600519" --domain finance --sub_domain finance.cn_stock --zone cn --max_results 5 python <anysearch>/scripts/anysearch_cli.py batch_search --queries '[{"query":"SEC EDGAR companyfacts API",

python <anysearch>/scripts/anysearch_cli.py extract "https://example.com/article"

Executable file

Executable file

Executable file

Executable file

帮我安装这个 skill：https://github.com/AlphaMao1/AlphaMao_Skills/tree/main/skills/serenity-thesis-tracker

<http://xhslink.com/o/4y5hTyS0hnf>

The script expects an existing PaiWork/CDP-compatible browser bridge at `http://localhost:3456` with these endpoints: `/health`, `/new`, `/eval`, `/scroll`, and `/close`. It does not install packages, launch Chrome, or bypass login. Chrome/

#!/usr/bin/env python3

DEFAULT_PROXY = "http://localhost:3456"

url = proxy_url.rstrip("/") + path     data = body.encode("utf-8") if body is not None else None     req = urllib.request.Request(url, data=data, method=method)

data = body.encode("utf-8") if body is not None else None     req = urllib.request.Request(url, data=data, method=method)     if body is not None:

req = urllib.request.Request(url, data=data, method=method)

req.add_header("Content-Type", "text/plain; charset=utf-8")     try:         with urllib.request.urlopen(req, timeout=timeout) as resp:

try:         with urllib.request.urlopen(req, timeout=timeout) as resp:             raw = resp.read().decode("utf-8")

with urllib.request.urlopen(req, timeout=timeout) as resp:

def open_timeline(proxy_url: str, handle: str) -> str:     url = "https://x.com/" + handle.lstrip("@")     path = "/new?url=" + urllib.parse.quote(url, safe="")

url = "https://x.com/" + handle.lstrip("@")     path = "/new?url=" + urllib.parse.quote(url, safe="")     payload = request_json(proxy_url, path, timeout=60)

f"# Serenity Daily Intake - {report_date.isoformat()}",         "",         f"- source: https://x.com/{handle.lstrip('@')}",

"",         f"- source: https://x.com/{handle.lstrip('@')}",         f"- fetched_at_utc: {utc_now().isoformat()}",

f"- source: https://x.com/{handle.lstrip('@')}",

The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),

and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).

帮我安装这个 skill：https://github.com/AlphaMao1/AlphaMao_Skills/tree/main/skills/stanford-vibe-coding-course

<http://xhslink.com/o/1qmfBbWYKku>

📚 本篇翻译提供了核心概念，但完整学习需要访问原链接： 🔗 https://github.com/modelcontextprotocol/servers

🔗 https://github.com/modelcontextprotocol/servers

### 📺 Slides (选读) - [Introduction and how an LLM is made](https://docs.google.com/presentation/d/1zT2Ofy88cajLTLkd7TcuSM4BCELvF9qQdHmlz33i4t0/edit?usp=sharing)

- [Introduction and how an LLM is made](https://docs.google.com/presentation/d/1zT2Ofy88cajLTLkd7TcuSM4BCELvF9qQdHmlz33i4t0/edit?usp=sharing)

- [Power prompting for LLMs](https://docs.google.com/presentation/d/1MIhw8p6TLGdbQ9TcxhXSs5BaPf5d_h77QY70RHNfeGs/edit?usp=drive_link)

### Assignment (可选) - [LLM Prompting Playground](https://github.com/mihail911/modern-software-dev-assignments/tree/master/week1) ⭐

- [LLM Prompting Playground](https://github.com/mihail911/modern-software-dev-assignments/tree/master/week1) ⭐

### 📺 Slides (选读) - [Building a coding agent from scratch](https://docs.google.com/presentation/d/11CP26VhsjnZOmi9YFgLlonzdib9BLyAlgc4cEvC5Fps/edit?usp=sharing) ⭐

- [Building a coding agent from scratch](https://docs.google.com/presentation/d/11CP26VhsjnZOmi9YFgLlonzdib9BLyAlgc4cEvC5Fps/edit?usp=sharing) ⭐

- [Building a custom MCP server](https://docs.google.com/presentation/d/1zSC2ra77XOUrJeyS85houg1DU7z9hq5Y4ebagTch-5o/edit?usp=drive_link)

### Assignment (可选) - [First Steps in the AI IDE](https://github.com/mihail911/modern-software-dev-assignments/tree/master/week2) 📖

- [First Steps in the AI IDE](https://github.com/mihail911/modern-software-dev-assignments/tree/master/week2) 📖

### 📺 Slides (选读) - [From first prompt to optimal IDE setup](https://docs.google.com/presentation/d/11pQNCde_mmRnImBat0Zymnp8TCS_cT_1up7zbcj6Sjg/edit?usp=sharing)

- [From first prompt to optimal IDE setup](https://docs.google.com/presentation/d/11pQNCde_mmRnImBat0Zymnp8TCS_cT_1up7zbcj6Sjg/edit?usp=sharing)

### Assignment (可选) - [Build a Custom MCP Server](https://github.com/mihail911/modern-software-dev-assignments/blob/master/week3/assignment.md) ⭐

- [Build a Custom MCP Server](https://github.com/mihail911/modern-software-dev-assignments/blob/master/week3/assignment.md) ⭐

### 📺 Slides (选读) - [How to be an agent manager](https://docs.google.com/presentation/d/19mgkwAnJDc7JuJy0zhhoY0ZC15DiNpxL8kchPDnRkRQ/edit?usp=sharing)

- [How to be an agent manager](https://docs.google.com/presentation/d/19mgkwAnJDc7JuJy0zhhoY0ZC15DiNpxL8kchPDnRkRQ/edit?usp=sharing)

### Assignment (可选) - [Coding with Claude Code](https://github.com/mihail911/modern-software-dev-assignments/blob/master/week4/assignment.md) ⭐

- [Coding with Claude Code](https://github.com/mihail911/modern-software-dev-assignments/blob/master/week4/assignment.md) ⭐

### 📺 Slides (选读) - [How to Build a Breakout AI Developer Product](https://docs.google.com/presentation/d/1Djd4eBLBbRkma8rFnJAWMT0ptct_UGB8hipmoqFVkxQ/edit?usp=sharing)

- [How to Build a Breakout AI Developer Product](https://docs.google.com/presentation/d/1Djd4eBLBbRkma8rFnJAWMT0ptct_UGB8hipmoqFVkxQ/edit?usp=sharing)

### Assignment (可选) - [Agentic Development with Warp](https://github.com/mihail911/modern-software-dev-assignments/tree/master/week5) 📖

- [Agentic Development with Warp](https://github.com/mihail911/modern-software-dev-assignments/tree/master/week5) 📖

### 📺 Slides (选读) - [AI QA, SAST, DAST, and Beyond](https://docs.google.com/presentation/d/1C05bCLasMDigBbkwdWbiz4WrXibzi6ua4hQQbTod_8c/edit?usp=sharing) ⭐

- [AI QA, SAST, DAST, and Beyond](https://docs.google.com/presentation/d/1C05bCLasMDigBbkwdWbiz4WrXibzi6ua4hQQbTod_8c/edit?usp=sharing) ⭐

### Assignment (可选) - [Writing Secure AI Code](https://github.com/mihail911/modern-software-dev-assignments/blob/master/week6/assignment.md) ⭐

- [Writing Secure AI Code](https://github.com/mihail911/modern-software-dev-assignments/blob/master/week6/assignment.md) ⭐

### 📺 Slides (选读) - [AI code review](https://docs.google.com/presentation/d/1NkPzpuSQt6Esbnr2-EnxM9007TL6ebSPFwITyVY-QxU/edit?usp=sharing)

- [AI code review](https://docs.google.com/presentation/d/1NkPzpuSQt6Esbnr2-EnxM9007TL6ebSPFwITyVY-QxU/edit?usp=sharing)

### Assignment (可选) - [Code Review Reps](https://github.com/mihail911/modern-software-dev-assignments/tree/master/week7) 📖

- [Code Review Reps](https://github.com/mihail911/modern-software-dev-assignments/tree/master/week7) 📖

### 📺 Slides (选读) - [Incident response and DevOps](https://docs.google.com/presentation/d/1Mfe-auWAsg9URCujneKnHr0AbO8O-_U4QXBVOlO4qp0/edit?usp=sharing)

- [Incident response and DevOps](https://docs.google.com/presentation/d/1Mfe-auWAsg9URCujneKnHr0AbO8O-_U4QXBVOlO4qp0/edit?usp=sharing)

使用 **Obsidian Spaced Repetition 插件**: - 插件地址: https://github.com/st3v3nmw/obsidian-spaced-repetition

- 插件地址: https://github.com/st3v3nmw/obsidian-spaced-repetition

--- 原文链接: https://www.youtube.com/watch?v=7xTGNNLPyMI

原文链接: https://www.youtube.com/watch?v=7xTGNNLPyMI

--- 原文链接: https://cloud.google.com/discover/what-is-prompt-engineering

原文链接: https://cloud.google.com/discover/what-is-prompt-engineering

--- 原文链接: https://www.promptingguide.ai/techniques

原文链接: https://www.promptingguide.ai/techniques

## 1. 零样本提示 (Zero-Shot Prompting) **原文链接**: [Zero-Shot Prompting](https://www.promptingguide.ai/techniques/zeroshot)

**原文链接**: [Zero-Shot Prompting](https://www.promptingguide.ai/techniques/zeroshot)

## 2. 少样本提示 (Few-Shot Prompting) **原文链接**: [Few-Shot Prompting](https://www.promptingguide.ai/techniques/fewshot)

**原文链接**: [Few-Shot Prompting](https://www.promptingguide.ai/techniques/fewshot)

## 3. 思维链提示 (Chain-of-Thought Prompting) **原文链接**: [Chain-of-Thought Prompting](https://www.promptingguide.ai/techniques/cot)

**原文链接**: [Chain-of-Thought Prompting](https://www.promptingguide.ai/techniques/cot)

## 4. 检索增强生成 (Retrieval Augmented Generation, RAG) **原文链接**: [Retrieval Augmented Generation](https://www.promptingguide.ai/techniques/rag)

**原文链接**: [Retrieval Augmented Generation](https://www.promptingguide.ai/techniques/rag)

## 5. ReAct 提示 **原文链接**: [ReAct Prompting](https://www.promptingguide.ai/techniques/react)

**原文链接**: [ReAct Prompting](https://www.promptingguide.ai/techniques/react)

## 6. 元提示 (Meta Prompting) **原文链接**: [Meta Prompting](https://www.promptingguide.ai/techniques/meta-prompting)

**原文链接**: [Meta Prompting](https://www.promptingguide.ai/techniques/meta-prompting)

## 7. 自洽性 (Self-Consistency) **原文链接**: [Self-Consistency](https://www.promptingguide.ai/techniques/consistency)

**原文链接**: [Self-Consistency](https://www.promptingguide.ai/techniques/consistency)

## 8. 生成知识提示 (Generate Knowledge Prompting) **原文链接**: [Generated Knowledge Prompting](https://www.promptingguide.ai/techniques/knowledge)

**原文链接**: [Generated Knowledge Prompting](https://www.promptingguide.ai/techniques/knowledge)

## 9. 提示链 (Prompt Chaining) **原文链接**: [Prompt Chaining](https://www.promptingguide.ai/techniques/prompt_chaining)

**原文链接**: [Prompt Chaining](https://www.promptingguide.ai/techniques/prompt_chaining)

## 10. 思维树 (Tree of Thoughts, ToT) **原文链接**: [Tree of Thoughts](https://www.promptingguide.ai/techniques/tot)

**原文链接**: [Tree of Thoughts](https://www.promptingguide.ai/techniques/tot)

## 11. 自动推理和工具使用 (ART) **原文链接**: [Automatic Reasoning and Tool-use](https://www.promptingguide.ai/techniques/art)

**原文链接**: [Automatic Reasoning and Tool-use](https://www.promptingguide.ai/techniques/art)

## 12. 自动提示工程师 (APE) **原文链接**: [Automatic Prompt Engineer](https://www.promptingguide.ai/techniques/ape)

**原文链接**: [Automatic Prompt Engineer](https://www.promptingguide.ai/techniques/ape)

## 13. 主动提示 (Active-Prompt) **原文链接**: [Active-Prompt](https://www.promptingguide.ai/techniques/activeprompt)

**原文链接**: [Active-Prompt](https://www.promptingguide.ai/techniques/activeprompt)

## 14. 定向刺激提示 (Directional Stimulus Prompting) **原文链接**: [Directional Stimulus Prompting](https://www.promptingguide.ai/techniques/dsp)

**原文链接**: [Directional Stimulus Prompting](https://www.promptingguide.ai/techniques/dsp)

## 15. 程序辅助语言模型 (PAL) **原文链接**: [Program-Aided Language Models](https://www.promptingguide.ai/techniques/pal)

**原文链接**: [Program-Aided Language Models](https://www.promptingguide.ai/techniques/pal)

## 16. Reflexion **原文链接**: [Reflexion](https://www.promptingguide.ai/techniques/reflexion)

**原文链接**: [Reflexion](https://www.promptingguide.ai/techniques/reflexion)

## 17. 多模态思维链 (Multimodal CoT) **原文链接**: [Multimodal CoT](https://www.promptingguide.ai/techniques/multimodalcot)

**原文链接**: [Multimodal CoT](https://www.promptingguide.ai/techniques/multimodalcot)

## 18. 图提示 (GraphPrompts) **原文链接**: [Graph Prompting](https://www.promptingguide.ai/techniques/graph)

**原文链接**: [Graph Prompting](https://www.promptingguide.ai/techniques/graph)

--- 原文链接: https://www.youtube.com/watch?v=T9aRN5JkmL8

原文链接: https://www.youtube.com/watch?v=T9aRN5JkmL8

--- 原文链接: https://cdn.openai.com/pdf/6a2631dc-783e-479b-b1a4-af0cfbd38630/how-openai-uses-codex.pdf

原文链接: https://cdn.openai.com/pdf/6a2631dc-783e-479b-b1a4-af0cfbd38630/how-openai-uses-codex.pdf

--- 原文链接: https://stytch.com/blog/model-context-protocol-introduction/

原文链接: https://stytch.com/blog/model-context-protocol-introduction/

## 什么是模型上下文协议 (MCP)? MCP本质上是 AI 应用程序与外部工具或数据源之间的通用适配器。它定义了一个[通用协议](https://www.anthropic.com/news/model-context-protocol)（建立在 JSON-RPC 2.0 之上），让 AI 助手以结构化的方式调用外部服务的函数、获取数据或使用预定义的提示。不再需要每个 LLM 应用程序为每个 API 或数据库编写自定义代码，[MCP 为所有交互提供了一种标准化的“语言”]

MCP本质上是 AI 应用程序与外部工具或数据源之间的通用适配器。它定义了一个[通用协议](https://www.anthropic.com/news/model-context-protocol)（建立在 JSON-RPC 2.0 之上），让 AI 助手以结构化的方式调用外部服务的函数、获取数据或使用预定义的提示。不再需要每个 LLM 应用程序为每个 API 或数据库编写自定义代码，[MCP 为所有交互提供了一种标准化的“语言”](https://medium.com/d

- **自主代理**: MCP 赋予了更自主的 AI 行为。代理不限于其内置知识；它们可以主动检索信息或在多步工作流中执行操作。例如，复杂的代理可能使用 MCP 从 CRM 收集数据，然后通过通信工具发送电子邮件，然后在数据库中记录一条记录——所有这些都在一个无缝链中完成。 - **减少摩擦和设置**: 由于 MCP 充当通用接口，开发人员避免了维护单独集成的碎片化。一旦应用程序支持 MCP，它就可以通过单一机制连接到任意数量的服务。这大大减少了每次你想让 AI 使用新 AP

- **减少摩擦和设置**: 由于 MCP 充当通用接口，开发人员避免了维护单独集成的碎片化。一旦应用程序支持 MCP，它就可以通过单一机制连接到任意数量的服务。这大大减少了每次你想让 AI 使用新 API 时所需的手动设置。 - **一致性和互操作性**: MCP 强制执行跨工具的一致请求/响应格式。这意味着你的 AI 应用程序不必处理服务 A 的 HTTP 响应、服务 B 的 XML 响应等。模型的输出（函数调用）和工具结果都以[统一的 JSON 结构](https://

- **一致性和互操作性**: MCP 强制执行跨工具的一致请求/响应格式。这意味着你的 AI 应用程序不必处理服务 A 的 HTTP 响应、服务 B 的 XML 响应等。模型的输出（函数调用）和工具结果都以[统一的 JSON 结构](https://dev.to/fotiecodes/function-calling-vs-model-context-protocol-mcp-what-you-need-to-know-4nbo)传递。这种一致性使得调试和扩展变得更加容易。

--- 原文链接: https://github.com/modelcontextprotocol/servers

原文链接: https://github.com/modelcontextprotocol/servers

此存储库是[模型上下文协议 (MCP)](https://modelcontextprotocol.io/) 的参考实现集合，以及社区构建的服务器和额外资源的参考。

> **重要提示** > > 如果你正在寻找 MCP 服务器列表，你可以在 [MCP Registry](https://registry.modelcontextprotocol.io/) 上浏览已发布的服务器。此存储库仅用于托管由 MCP 指导小组维护的少量参考服务器。

> 如果你正在寻找 MCP 服务器列表，你可以在 [MCP Registry](https://registry.modelcontextprotocol.io/) 上浏览已发布的服务器。此存储库仅用于托管由 MCP 指导小组维护的少量参考服务器。

- **[Everything](https://github.com/modelcontextprotocol/servers/blob/main/src/everything)**: 包含提示、资源和工具的参考/测试服务器。 - **[Fetch](https://github.com/modelcontextprotocol/servers/blob/main/src/fetch)**: 用于高效 LLM 使用的 Web 内容获取和转换。

- **[Fetch](https://github.com/modelcontextprotocol/servers/blob/main/src/fetch)**: 用于高效 LLM 使用的 Web 内容获取和转换。

- **[Filesystem](https://github.com/modelcontextprotocol/servers/blob/main/src/filesystem)**: 具有可配置访问控制的安全文件操作。

- **[Git](https://github.com/modelcontextprotocol/servers/blob/main/src/git)**: 读取、搜索和操作 Git 存储库的工具。

- **[Memory](https://github.com/modelcontextprotocol/servers/blob/main/src/memory)**: 基于知识图谱的持久记忆系统。

- **[Sequential Thinking](https://github.com/modelcontextprotocol/servers/blob/main/src/sequentialthinking)**: 通过思维序列进行动态和反思性的问题解决。

- **[Time](https://github.com/modelcontextprotocol/servers/blob/main/src/time)**: 时间和时区转换功能。

- 搜索文件 - 获取文件元数据 - 通过 [Roots](https://modelcontextprotocol.io/docs/learn/client-concepts#roots) 进行动态目录访问控制

- 通过 [Roots](https://modelcontextprotocol.io/docs/learn/client-concepts#roots) 进行动态目录访问控制

例如，這将启动 [Memory](https://github.com/modelcontextprotocol/servers/blob/main/src/memory) 服务器：

例如，這将启动 [Git](https://github.com/modelcontextprotocol/servers/blob/main/src/git) 服务器：

[Credential-related source omitted]

- **PostgreSQL**: 可以只读访问 PostgreSQL 数据库并运行 SQL。 - **Slack**: 用于与 Slack 交互的服务器。 ... (更多请参阅[完整列表](https://github.com/modelcontextprotocol/servers))

... (更多请参阅[完整列表](https://github.com/modelcontextprotocol/servers))

--- 原文链接: https://developers.cloudflare.com/agents/guides/remote-mcp-server/#add-authentication

原文链接: https://developers.cloudflare.com/agents/guides/remote-mcp-server/#add-authentication

在此示例中，我们使用 GitHub 作为[OAuth 提供者](https://developers.cloudflare.com/agents/model-context-protocol/authorization/#2-third-party-oauth-provider)，但你可以将 MCP 服务器与任何支持 OAuth 2.0 规范的 OAuth 提供者连接，包括 Google、Slack、Stytch、Auth0、WorkOS 等。

你需要创建两个 [GitHub OAuth App ↗](https://docs.github.com/en/apps/oauth-apps/building-oauth-apps/creating-an-oauth-app) 才能使用 GitHub 作为 MCP 服务器的身份验证提供者——一个用于本地开发，一个用于生产。

### 首先为本地开发创建一个新的 OAuth App 导航到 [github.com/settings/developers ↗](https://github.com/settings/developers) 创建一个新的 OAuth App，设置如下：

导航到 [github.com/settings/developers ↗](https://github.com/settings/developers) 创建一个新的 OAuth App，设置如下：

- **Application name**: My MCP Server (local) - **Homepage URL**: `http://localhost:8788` - **Authorization callback URL**: `http://localhost:8788/callback`

- **Homepage URL**: `http://localhost:8788`

- **Authorization callback URL**: `http://localhost:8788/callback`

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

你的 MCP 服务器现在运行在 `http://localhost:8788/mcp`。

在一个新终端中，运行 [MCP Inspector ↗](https://github.com/modelcontextprotocol/inspector)。MCP Inspector 是一个交互式 MCP 客户端，允许你连接到 MCP 服务器并从 Web 浏览器调用工具。

在 Web 浏览器中打开 MCP Inspector（通常是 `http://localhost:5173`）。

在 Inspector 中，输入你的 MCP 服务器的 URL `http://localhost:8788/mcp`。

### 其次 — 为生产环境创建一个新的 OAuth App 你需要重复这些步骤为生产环境创建一个新的 OAuth App。 导航到 [github.com/settings/developers ↗](https://github.com/settings/developers) 创建一个新的 OAuth App，设置如下：

你需要重复这些步骤为生产环境创建一个新的 OAuth App。 导航到 [github.com/settings/developers ↗](https://github.com/settings/developers) 创建一个新的 OAuth App，设置如下： - **Application name**: My MCP Server (production)

导航到 [github.com/settings/developers ↗](https://github.com/settings/developers) 创建一个新的 OAuth App，设置如下：

[Credential-related source omitted]

--- 原文链接: https://github.com/modelcontextprotocol/typescript-sdk

原文链接: https://github.com/modelcontextprotocol/typescript-sdk

> 我们预计在 2026 年第一季度发布稳定的 v2 版本。在此之前，**v1.x 仍然是生产使用的推荐版本**。在 v2 发布后，v1.x 将继续接收至少 6 个月的错误修复和安全更新，以便人们有时间升级。 > > 有关 v1 文档和代码，请参阅 [v1.x 分支](https://github.com/modelcontextprotocol/typescript-sdk/tree/v1.x)。

> 有关 v1 文档和代码，请参阅 [v1.x 分支](https://github.com/modelcontextprotocol/typescript-sdk/tree/v1.x)。

--- 原文链接: https://blog.modelcontextprotocol.io/posts/2025-09-08-mcp-registry-preview/

原文链接: https://blog.modelcontextprotocol.io/posts/2025-09-08-mcp-registry-preview/

MCP Registry 现在提供**预览版**。要开始使用： - **服务器维护者**: 按照我们的指南[将服务器添加到 MCP Registry](https://github.com/modelcontextprotocol/registry/blob/main/docs/modelcontextprotocol-io/quickstart.mdx)。

- **服务器维护者**: 按照我们的指南[将服务器添加到 MCP Registry](https://github.com/modelcontextprotocol/registry/blob/main/docs/modelcontextprotocol-io/quickstart.mdx)。

- **客户端维护者**: 按照我们的指南[访问 MCP Registry 数据](https://github.com/modelcontextprotocol/registry/blob/main/docs/modelcontextprotocol-io/registry-aggregators.mdx)。

在 2025 年 3 月，我们分享了我们想要为 MCP 生态系统建立一个中央注册表的想法。今天我们宣布，我们已经推出了 [registry.modelcontextprotocol.io](https://registry.modelcontextprotocol.io) 作为官方的 MCP Registry。作为 MCP 项目的一部分，MCP Registry 以及父级 [OpenAPI 规范](https://github.com/modelcontextprotocol

在构建中央注册表时，对我们来说重要的是不要夺走社区和公司已经建立的现有注册表。MCP Registry 作为公开可用 MCP 服务器的主要事实来源，组织可以选择基于自定义标准[创建子注册表](https://github.com/modelcontextprotocol/registry/blob/main/docs/modelcontextprotocol-io/registry-aggregators.mdx)。例如：

MCP Registry 是一个由注册表工作组维护并获得许可的官方 MCP 项目。社区成员可以提交 issue 来标记违反 MCP [审核指南](https://github.com/modelcontextprotocol/registry/blob/main/docs/modelcontextprotocol-io/moderation-policy.mdx)的服务器——例如包含垃圾邮件、恶意代码或冒充合法服务的服务器。注册表维护者随后可以将这些条目列入黑名单，并追溯性地

--- 原文链接: https://www.reillywood.com/blog/apis-dont-make-good-mcp-tools/

原文链接: https://www.reillywood.com/blog/apis-dont-make-good-mcp-tools/

[模型上下文协议](https://modelcontextprotocol.io/overview) (MCP) 如今是一件大事。它已成为让 LLM 访问他人编写的工具的事实标准，这当然会将它们转变为[代理](https://simonwillison.net/2025/May/22/tools-in-a-loop/)。但是为新的 MCP 服务器编写工具很难，因此人们经常建议[将现有 API 自动转换为 MCP 工具](https://blog.christianposta

众所周知，[VS Code 有 128 个工具的硬性限制](https://code.visualstudio.com/docs/copilot/chat/chat-agent-mode) - 但[许多模型在该数量之前就难以准确调用工具](https://arxiv.org/abs/2411.15399)。此外，每个工具及其描述都占据了宝贵的上下文窗口空间。

从头开始设计的 MCP 工具通常比单个 Web API [灵活得多](https://engineering.block.xyz/blog/blocks-playbook-for-designing-mcp-servers)，每个工具都能够完成多个单独 API 的工作。

API 通常按记录数分页，但记录的大小差异很大。一条记录可能包含一个占用 100,000 个 [token](https://learn.microsoft.com/en-us/dotnet/ai/conceptual/understanding-tokens) 的大文本字段，而另一条可能只包含 10 个。将这些 API 结果直接放入代理的上下文窗口就像一场赌博；有时能行，有时会爆炸。

数据的格式也可能是一个问题。如今大多数 Web API 返回 JSON，但 JSON 是一种非常低效的 token 格式。 对比 JSON 和 CSV：CSV 数据更简洁 - 每条记录消耗的 token 只有一半。[通常 CSV, TSV 或 YAML (对于嵌套数据) 是比 JSON 更好的选择](https://david-gilbertson.medium.com/llm-output-formats-why-json-costs-more-than-tsv-ebaf5

对比 JSON 和 CSV：CSV 数据更简洁 - 每条记录消耗的 token 只有一半。[通常 CSV, TSV 或 YAML (对于嵌套数据) 是比 JSON 更好的选择](https://david-gilbertson.medium.com/llm-output-formats-why-json-costs-more-than-tsv-ebaf590bd541)。

这些问题都不是不可克服的。你可以想象自动添加工具参数让代理[投影](https://en.wikipedia.org/wiki/Projection_(relational_algebra))字段，自动截断或总结大结果，以及自动将 JSON 结果转换为 CSV。但我见过的大多数服务器都没有做这些事情。

这种分层和工具链在 MCP 服务器中[非常有效](https://engineering.block.xyz/blog/build-mcp-tools-like-ogres-with-layers)，如果自动将 API 转换为工具，你会完全错过这一点。

像 Claude Code 这样的代理如今非常擅长编写+执行代码，包括调用 Web API 的脚本。有些人甚至[认为根本不需要 MCP](https://lucumr.pocoo.org/2025/7/3/tools/)！

我不同意这个结论，但我确实认为我们应该顺势而为。[代理的沙盒化正在迅速改进](https://github.com/openai/codex)，如果代理直接调用 API 既简单又安全，那么我们不妨这样做，省去中间人。

--- 原文链接: https://blog.ravi-mehta.com/p/specs-are-the-new-source-code

原文链接: https://blog.ravi-mehta.com/p/specs-are-the-new-source-code

## 为什么 PM 突然成了瓶颈 Andrew Ng 在他最近的演讲 [Building Faster with AI](https://www.youtube.com/watch?v=RNJCfif1dPY) 中指出了一个前所未有的趋势：

Andrew Ng 在他最近的演讲 [Building Faster with AI](https://www.youtube.com/watch?v=RNJCfif1dPY) 中指出了一个前所未有的趋势：

OpenAI 的 Sean Grove 有一个挑衅性的论点。在他最近的演讲 [The New Code](https://www.youtube.com/watch?v=8rABwKRsec4) 中，他认为**写得好的 Prompt（即规范）是新的源代码**。

--- 原文链接: https://www.dbreunig.com/2025/06/22/how-contexts-fail-and-how-to-fix-them.html

原文链接: https://www.dbreunig.com/2025/06/22/how-contexts-fail-and-how-to-fix-them.html

DeepMind 团队在 [Gemini 2.5 技术报告](https://storage.googleapis.com/deepmind-media/gemini/gemini_v2_5_report.pdf)中指出了上下文中毒问题。在玩 Pokémon 时，Gemini Agent 偶尔会在游戏中产生幻觉，从而毒害其上下文：

对于较小的模型，分心的上限要低得多。一项 [Databricks 研究](https://www.databricks.com/blog/long-context-rag-performance-llms)发现，Llama 3.1 405b 的模型正确性在 32k左右开始下降，而较小的模型则更早。

有一段时间，似乎每个人都会发布一个 [MCP](https://modelcontextprotocol.io/)。一个强大的模型，连接到你所有的服务和东西，做你所有琐碎任务的梦想似乎触手可及。只需将所有工具描述扔进 Prompt 并点击运行。[Claude 的系统提示词](https://docs.anthropic.com/en/release-notes/system-prompts)向我们展示了这种方式，因为它主要是工具定义或使用工具的说明。

[伯克利函数调用排行榜 (Berkeley Function-Calling Leaderboard)](https://gorilla.cs.berkeley.edu/leaderboard.html) 是一个评估模型有效使用工具能力的基准。现在是第 3 版，排行榜显示，当提供多个工具时，每个模型的表现都会变差。此外，伯克利团队设计了一些场景，“其中提供的函数都不相关……我们期望模型的输出是没有函数调用。”然而，所有模型偶尔都会调用不相关的工具。

[最近的一篇论文](https://arxiv.org/pdf/2411.15399)中可以看到上下文困惑的一个惊人例子，该论文评估了小模型在 [GeoEngine 基准](https://arxiv.org/abs/2404.15500)上的表现，该试验包含 46 种不同的工具。当团队给量化（压缩）的 Llama 3.1 8b 一个包含所有 46 个工具的查询时，它失败了，即使上下文完全在 16k 上下文窗口内。但当他们只给模型 19 个工具时，它成功了。

Microsoft 和 Salesforce 团队在[最近的一篇论文](https://arxiv.org/pdf/2505.06120)中精彩地记录了这一点。该团队从多个基准中提取 Prompt，并将其信息“分片”到多个 Prompt 中。可以这样想：有时，你可能会坐下来在 ChatGPT 或 Claude 中输入段落，然后在点击回车之前考虑每个必要的细节。其他时候，你可能会从一个简单的 Prompt 开始，然后在聊天机器人的回答不令人满意时添加更多细节。Microsoft

请阅读后续文章：“[如何修复你的上下文](https://www.dbreunig.com/2025/06/26/how-to-fix-your-context.html)”

--- 原文链接: https://devin.ai/agents101#introduction

原文链接: https://devin.ai/agents101#introduction

--- 原文链接: https://github.com/humanlayer/advanced-context-engineering-for-coding-agents/blob/main/ace-fca.md

原文链接: https://github.com/humanlayer/advanced-context-engineering-for-coding-agents/blob/main/ace-fca.md

众所周知，AI 编码工具在真实的生产代码库中表现挣扎。[斯坦福关于 AI 对开发者生产力影响的研究](https://www.youtube.com/watch?v=tbDDYKRFjhk)发现：

你也可以[使用提交信息进行有意压缩](https://x.com/dexhorthy/status/1961490837017088051)。

--- 原文链接: https://www.anthropic.com/engineering/writing-tools-for-agents

原文链接: https://www.anthropic.com/engineering/writing-tools-for-agents

在计算中，确定性系统（如函数 `getWeather("NYC")`）每次给定相同的输入都会产生相同的输出。 工具反映了确定性系统与**非确定性 Agent** 之间的契约。当用户问“我今天应该带伞吗？”时，Agent 可能会调用天气工具，也可能根据常识回答，或者先问地点。 这意味着我们需要从根本上重新思考软件编写方式：不是像为其他开发者编写 API 那样编写工具和 [MCP 服务器](https://modelcontextprotocol.io/)，而是需要**为 Age

这意味着我们需要从根本上重新思考软件编写方式：不是像为其他开发者编写 API 那样编写工具和 [MCP 服务器](https://modelcontextprotocol.io/)，而是需要**为 Agent 设计它们**。

### 1. 构建原型 (Building a prototype) 如果不亲自动手，很难预料哪些工具对 Agent 来说是顺手的。 - **使用 Claude Code**: 为 Claude 提供你工具依赖的库、API 或 SDK 的文档。LLM 友好的文档通常可以在 `llms.txt` 中找到（例如 [Anthropic 的 llms.txt](https://docs.anthropic.com/llms.txt)）。

如果不亲自动手，很难预料哪些工具对 Agent 来说是顺手的。 - **使用 Claude Code**: 为 Claude 提供你工具依赖的库、API 或 SDK 的文档。LLM 友好的文档通常可以在 `llms.txt` 中找到（例如 [Anthropic 的 llms.txt](https://docs.anthropic.com/llms.txt)）。 - **本地测试**: 将工具包装在本地 MCP 服务器中，使用 `claude mcp add <name> <c

- **使用 Claude Code**: 为 Claude 提供你工具依赖的库、API 或 SDK 的文档。LLM 友好的文档通常可以在 `llms.txt` 中找到（例如 [Anthropic 的 llms.txt](https://docs.anthropic.com/llms.txt)）。

--- 原文链接: https://www.anthropic.com (Internal PDF)

原文链接: https://www.anthropic.com (Internal PDF)

--- 原文链接: https://code.claude.com/docs/en/best-practices

原文链接: https://code.claude.com/docs/en/best-practices

- **Allowlists**: 允许已知的安全工具（如 `npm run lint`）自动运行。 - **Sandboxing**: 启用 `/sandbox` 进行操作系统级隔离，限制文件系统和网络访问。 - **Safe Autonomous Mode**: 使用 `--dangerously-skip-permissions` 进行全自动运行（**警告**: 仅在沙箱或无关紧要的环境中使用）。

- **Safe Autonomous Mode**: 使用 `--dangerously-skip-permissions` 进行全自动运行（**警告**: 仅在沙箱或无关紧要的环境中使用）。

--- 原文链接: https://github.com/vijaythecoder/awesome-claude-agents

原文链接: https://github.com/vijaythecoder/awesome-claude-agents

--- 原文链接: https://github.com/SuperClaude-Org/SuperClaude_Framework

原文链接: https://github.com/SuperClaude-Org/SuperClaude_Framework

--- 原文链接: https://blog.stockapp.com/good-context-good-code/

原文链接: https://blog.stockapp.com/good-context-good-code/

--- 原文链接: https://medium.com/@OutSightAI/peeking-under-the-hood-of-claude-code-be9182046875

原文链接: https://medium.com/@OutSightAI/peeking-under-the-hood-of-claude-code-be9182046875

--- 原文链接: https://www.warp.dev/warp-university

原文链接: https://www.warp.dev/warp-university

--- 原文链接: https://warp.dev/blog/warp-ai-vs-claude-code

原文链接: https://warp.dev/blog/warp-ai-vs-claude-code

--- 原文链接: https://www.warp.dev/blog/how-we-use-warp-to-build-warp

原文链接: https://www.warp.dev/blog/how-we-use-warp-to-build-warp

--- 原文链接: https://www.splunk.com/en_us/blog/learn/sast-vs-dast.html

原文链接: https://www.splunk.com/en_us/blog/learn/sast-vs-dast.html

*   高亮显示易受攻击或已弃用的函数的使用，例如 JavaScript 中的 `eval()` 或 C 中的 `strcpy()`。

### 何时使用 SAST *   在早期开发阶段：一旦编写了代码就实施 SAST。这有助于在它们深深嵌入代码库之前捕获漏洞。 *   为了合规性：许多行业法规（如 [PCI-DSS](https://www.splunk.com/en_us/blog/learn/pci-dss-compliance.html) 和 HIPAA）要求对源代码进行安全审查。

*   在早期开发阶段：一旦编写了代码就实施 SAST。这有助于在它们深深嵌入代码库之前捕获漏洞。 *   为了合规性：许多行业法规（如 [PCI-DSS](https://www.splunk.com/en_us/blog/learn/pci-dss-compliance.html) 和 HIPAA）要求对源代码进行安全审查。 *   在 CI/CD 管道中：自动化 SAST 扫描以确保持续安全性，而不会减慢开发速度。

*   为了合规性：许多行业法规（如 [PCI-DSS](https://www.splunk.com/en_us/blog/learn/pci-dss-compliance.html) 和 HIPAA）要求对源代码进行安全审查。

*   对于遗留应用程序：当源代码不可用或过时时，DAST 是识别漏洞的唯一方法。 *   验证补丁：使用 DAST 验证安全修复是否有效并且没有引入新的问题。 *   为了渗透测试：DAST 模拟外部攻击者的视角，使其成为[渗透测试](https://www.splunk.com/en_us/blog/learn/penetration-testing.html)的重要工具。

*   为了渗透测试：DAST 模拟外部攻击者的视角，使其成为[渗透测试](https://www.splunk.com/en_us/blog/learn/penetration-testing.html)的重要工具。

--- 原文链接: https://embracethered.com/blog/posts/2025/github-copilot-remote-code-execution-via-prompt-injection/

原文链接: https://embracethered.com/blog/posts/2025/github-copilot-remote-code-execution-via-prompt-injection/

## 负责任的披露 (Responsible Disclosure) 在 2025 年 6 月 29 日报告该漏洞后，Microsoft 确认了复现并问了几个后续问题。几周后，MSRC 指出这是他们已经在跟踪的问题，并且将在 8 月修复。随着 8 月补丁星期二的发布，此问题现已修复。 向来自 [Persistent Security](https://persistent-security.net/) 的 [Markus Vervier](https://x.com/marv

在 2025 年 6 月 29 日报告该漏洞后，Microsoft 确认了复现并问了几个后续问题。几周后，MSRC 指出这是他们已经在跟踪的问题，并且将在 8 月修复。随着 8 月补丁星期二的发布，此问题现已修复。 向来自 [Persistent Security](https://persistent-security.net/) 的 [Markus Vervier](https://x.com/marver) 致敬，他也识别并向 Microsoft 报告了这个漏洞。你可以

向来自 [Persistent Security](https://persistent-security.net/) 的 [Markus Vervier](https://x.com/marver) 致敬，他也识别并向 Microsoft 报告了这个漏洞。你可以在[这里](https://www.persistent-security.net/post/part-iii-vscode-copilot-wormable-command-execution-via-prompt

--- 原文链接: https://semgrep.dev/blog/2025/finding-vulnerabilities-in-modern-web-apps-using-claude-code-and-openai-codex/

原文链接: https://semgrep.dev/blog/2025/finding-vulnerabilities-in-modern-web-apps-using-claude-code-and-openai-codex/

--- 原文链接: https://unit42.paloaltonetworks.com/agentic-ai-threats/

原文链接: https://unit42.paloaltonetworks.com/agentic-ai-threats/

## AI 代理的安全风险 (Security Risks of AI Agents) 由于 AI 代理通常建立在 LLM 之上，它们继承了 [OWASP Top 10 for LLMs](https://owasp.org/www-project-top-10-for-large-language-model-applications/) 中概述的许多安全风险，例如提示注入、敏感数据泄露和供应链漏洞。然而，AI 代理通过集成通常以各种编程语言和框架构建的外部工具，超越了传统

由于 AI 代理通常建立在 LLM 之上，它们继承了 [OWASP Top 10 for LLMs](https://owasp.org/www-project-top-10-for-large-language-model-applications/) 中概述的许多安全风险，例如提示注入、敏感数据泄露和供应链漏洞。然而，AI 代理通过集成通常以各种编程语言和框架构建的外部工具，超越了传统的 LLM 应用程序。

--- 原文链接: https://owasp.org/www-project-top-10-for-large-language-model-applications/

原文链接: https://owasp.org/www-project-top-10-for-large-language-model-applications/

--- 原文链接: https://research.trychroma.com/context-rot

原文链接: https://research.trychroma.com/context-rot

--- 原文链接: https://github.com/SeanHeelan/o3_finds_cve-2025-37899/blob/master/system_prompt_uafs.prompt

原文链接: https://github.com/SeanHeelan/o3_finds_cve-2025-37899/blob/master/system_prompt_uafs.prompt

> **Original Link**: [https://blog.codinghorror.com/code-reviews-just-do-it/](https://blog.codinghorror.com/code-reviews-just-do-it/)

在 [Humanizing Peer Reviews](https://web.archive.org/web/20060315135514/http://www.processimpact.com/articles/humanizing_reviews.html) 一文中，Karl Wiegers 以一句强有力的断言开篇：

但不要只听我的一面之词。McConnell 在 [Code Complete](http://www.amazon.com/exec/obidos/ASIN/0735619670) 中提供了大量证据证明代码审查的有效性：

如果你的组织刚开始接触代码审查，我强烈推荐 Karl 的书 [Peer Reviews in Software](http://www.amazon.com/exec/obidos/ASIN/0201734850): A Practical Guide。Karl 在他的网站上提供的 [样章](https://web.archive.org/web/20060315135046/http://www.processimpact.com/reviews_book/reviews_

> **Original Link**: [https://github.blog/developer-skills/github/how-to-review-code-effectively-a-github-staff-engineers-philosophy/](https://github.blog/developer-skills/github/how-to-review-code-effectively-a-github-staff-engineers-philo

严格来说，代码审查——通过 GitHub 上的 [拉取请求审查](https://docs.github.com/pull-requests/collaborating-with-pull-requests/reviewing-changes-in-pull-requests/about-pull-request-reviews)——允许协作者对 PR 中提议的变更进行评论，表示对变更的批准，或在 PR 合并前要求进一步的变更。

我生活在我的 [GitHub 通知收件箱](https://github.com/notifications?query=is:unread) 中。这是我在浏览器中固定的少数几个标签页之一，即使它总是可用的。每当我等待 CI，在任务之间，开始新的一天，或者通常有空闲时刻时，我都喜欢检查我的收件箱。我在那里发现了大部分我也审查的 PR。GitHub 的团队倾向于拥有一个特定的 Slack 频道作为大本营，这也是分享准备好审查的 PR 的好地方——这是我发现 PR 的另一个主要方

我也很幸运地使用 [GitHub Slack 集成](https://slack.github.com/) 订阅 Slack 频道，以获取与我的团队相关的特定 PR。为了过滤哪些 PR 显示在 Slack 中，我使用特定于团队的标签，然后使用 Slack 中的 `subscribe` 命令，如 `/github subscribe your/repo pulls +label:"your-team-label"`。

你不希望代码变更去 ping 一个如此大的团队，以至于团队中的每个人都认为审查变更不是他们的 [责任](https://en.wikipedia.org/wiki/Diffusion_of_responsibility)（责任分散）。这可能导致 PR 要么被搁置未审，要么在不该合并时被合并，因为关键审查者在大量的通知中错过了它们。这两种情况都会影响产品的质量。

我建议如果可能的话，精简你所在的 [代码所有者 (code owner)](https://docs.github.com/repositories/managing-your-repositorys-settings-and-features/customizing-your-repository/about-code-owners) 团队的数量，以保持通知的可管理性。这样，落入你收件箱的 PR 就不再只是噪音，而是你真正觉得应该审查的东西。虽然大的、包罗万象的代码所有者团

另一种限制基于团队的通知的方法是创建一个“第一响应者”团队，然后使用自动化根据时间表添加和移除团队成员。这可以让你的团队专注于他们日常的代码库，而预定的第一响应者则会收到你团队服务领域内 PR 的通知。例如，使用 [PagerDuty API](https://developer.pagerduty.com/api-reference/3f03afb2c84a4-get-a-schedule)，你可以确定某一天谁是第一响应者。然后你可以使用 [Octokit 库](https

仓库级配置和自动化，例如使用 [CODEOWNERS 文件](https://docs.github.com/repositories/managing-your-repositorys-settings-and-features/customizing-your-repository/about-code-owners#codeowners-file-location) 和 [分支保护规则](https://docs.github.com/repositories/conf

我很推崇编写测试，因为它们消除了部分偏见。当你编写测试来检查代码是否正常工作时，[你不需要听作者的一面之词](https://www.youtube.com/watch?v=NIKAsGC1Iy8)，你只需要看测试是否通过——当然，前提是你把测试写对了。😅

> **Tip**: 使用 [特性标志 (feature flags)](https://github.com/flippercloud/flipper) 并使你的 PR 变小。你的更改越不吓人，它在生产中被逆转得越快，我就越容易批准它！

[‘Request changes’ 选项](https://docs.github.com/pull-requests/collaborating-with-pull-requests/reviewing-changes-in-pull-requests/reviewing-proposed-changes-in-a-pull-request#submitting-your-review) 会阻止 PR 合并，直到审查者回来并批准它。我很少使用它，因为它通常感觉太严厉了。我

GitHub 高级软件工程师 [Paul Smith](https://github.com/paulcsmith) 教会在请别人审查之前先审查我自己的 PR，我建议你也这样做。过一遍并在不明显的更改或如果在别人的 PR 中看到你会问的地方留下内联评论。自我审查也可以帮助确定 PR 是否太大，是否受益于 [被拆分](https://github.blog/2020-05-21-github-protips-tips-tricks-hacks-and-secrets-from-

> **Shout-out**: 如果你关心保持你的 PR 小，你可以使用 [lerebear/sizeup-action](https://github.com/lerebear/sizeup-action) 来自动给 PR 打标签，表明它们的复杂性和大小。

当你创建一个新的 PR 时，你可以选择将 PR 设置为草稿。我非常依赖 [草稿阶段](https://docs.github.com/pull-requests/collaborating-with-pull-requests/proposing-changes-to-your-work-with-pull-requests/changing-the-stage-of-a-pull-request) 来表明我是否想要审查。例如，如果所需的 CI 构建失败了或者我还没完成，我会

> **Original Link**: [https://arxiv.org/abs/2405.13565](https://arxiv.org/abs/2405.13565)

INSERT 153 COMMENT https://go.dev/doc/comment#func

> **Original Link**: [https://graphite.com/guides/ai-code-review-implementation-best-practices](https://graphite.com/guides/ai-code-review-implementation-best-practices)

随着人工智能越来越多地融入软件开发工作流程，[AI 代码审查](https://graphite.com/guides/how-ai-code-review-works)已成为提高代码质量和开发人员生产力的有用工具。本技术指南探讨了 AI 代码审查系统的实施，并概述了在开发过程中有效利用这些工具的最佳实践。

### 步骤 1：选择合适的工具 有几种 AI 代码审查工具可供选择，功能各异： - [Graphite Agent](https://graphite.com/features/agent)：通过上下文代码分析和 PR 自动化提供即时、可操作的反馈。

有几种 AI 代码审查工具可供选择，功能各异： - [Graphite Agent](https://graphite.com/features/agent)：通过上下文代码分析和 PR 自动化提供即时、可操作的反馈。 - [GitHub Copilot](https://github.com/features/copilot)：在编码时提供实时建议，以便为审查提供更清晰的代码。

- [Graphite Agent](https://graphite.com/features/agent)：通过上下文代码分析和 PR 自动化提供即时、可操作的反馈。

- [GitHub Copilot](https://github.com/features/copilot)：在编码时提供实时建议，以便为审查提供更清晰的代码。

- [SonarQube with AI](https://www.sonarsource.com/products/sonarqube/)：结合了传统静态分析和 AI 能力。

- [DeepCode](https://snyk.io/platform/deepcode-ai/)：专注于检测安全漏洞。

> **Original Link**: [https://blakesmith.me/2015/02/09/code-review-essentials-for-software-teams.html](https://blakesmith.me/2015/02/09/code-review-essentials-for-software-teams.html)

原文链接: https://www.youtube.com/watch?v=TswQeKftnaw

> **Original Link**: [https://sre.google/sre-book/introduction/](https://sre.google/sre-book/introduction/)

希望系统高度可靠是理所当然的，但这并不意味着冲突也不可避免。Google 选择了一种不同的方式来运行我们的系统：我们的 [站点可靠性工程 (SRE) 团队](https://sre.google/sre-book/software-engineering-in-sre/) 专注于雇佣软件工程师来运行我们的产品，并 [创建系统来完成那些工作](https://sre.google/sre-book/distributed-periodic-scheduling/)——那些通常由

SRE 做的根本工作在历史上一直由运维团队完成，但使用的是具有软件专业知识的工程师。我们依靠这样一个事实：这些工程师在本质上通过设计和 [实施自动化](https://sre.google/sre-book/automation-at-google/) 来取代人工劳动。

## DevOps 还是 SRE？(DevOps or SRE?) “DevOps”一词出现于 2008 年底。其核心原则——IT 职能在系统设计和开发的每个阶段的参与、严重依赖自动化而非人力、将工程实践和工具应用于运维任务——与许多 [SRE 的原则和实践](https://sre.google/sre-book/part-II-principles/) 是一致的。可以将 DevOps 视为将几个核心 SRE 原则推广到更广泛的组织、管理结构和人员。同样也可以将 SRE 视

“DevOps”一词出现于 2008 年底。其核心原则——IT 职能在系统设计和开发的每个阶段的参与、严重依赖自动化而非人力、将工程实践和工具应用于运维任务——与许多 [SRE 的原则和实践](https://sre.google/sre-book/part-II-principles/) 是一致的。可以将 DevOps 视为将几个核心 SRE 原则推广到更广泛的组织、管理结构和人员。同样也可以将 SRE 视为 DevOps 的一种特定实现，并带有一些独特的扩展。

### 确保对工程的持久关注 (Ensuring a Durable Focus on Engineering) 如前所述，Google 将 SRE 的运维工作上限设定为其时间的 50%。这在实践中是通过监控运维工作量来实现的，并将 [多余的运维工作](https://sre.google/sre-book/dealing-with-interrupts/) 重定向回产品开发团队。

如前所述，Google 将 SRE 的运维工作上限设定为其时间的 50%。这在实践中是通过监控运维工作量来实现的，并将 [多余的运维工作](https://sre.google/sre-book/dealing-with-interrupts/) 重定向回产品开发团队。

### 在不违反服务 SLO 的情况下追求最大变更速度 (Pursuing Maximum Change Velocity Without Violating a Service’s SLO) 产品开发和 SRE 团队可以通过消除各自目标中的结构性冲突来享受 [富有成效的工作关系](https://sre.google/workbook/engagement-model/)。这种冲突在于创新速度和产品稳定性之间。我们引入 **错误预算 (Error Budget)** 来解决

产品开发和 SRE 团队可以通过消除各自目标中的结构性冲突来享受 [富有成效的工作关系](https://sre.google/workbook/engagement-model/)。这种冲突在于创新速度和产品稳定性之间。我们引入 **错误预算 (Error Budget)** 来解决这个问题。

> **Original Link**: [https://last9.io/blog/traces-spans-observability-basics/](https://last9.io/blog/traces-spans-observability-basics/)

### OpenTelemetry：行业标准 [OpenTelemetry](https://opentelemetry.io/) 已成为实现 Trace 和 Span 的首选框架。它提供所有主要编程语言的库、供应商中立的 API 和 SDK，以及一致的数据收集和导出方式。

[OpenTelemetry](https://opentelemetry.io/) 已成为实现 Trace 和 Span 的首选框架。它提供所有主要编程语言的库、供应商中立的 API 和 SDK，以及一致的数据收集和导出方式。

> **Original Link**: [https://resolve.ai/blog/kubernetes-troubleshooting-in-resolve-ai](https://resolve.ai/blog/kubernetes-troubleshooting-in-resolve-ai)

> **Original Link**: [https://resolve.ai/blog/product-deep-dive](https://resolve.ai/blog/product-deep-dive)

> **Original Link**: [https://resolve.ai/blog/role-of-multi-agent-systems-AI-native-engineering](https://resolve.ai/blog/role-of-multi-agent-systems-AI-native-engineering)

> **Original Link**: [https://resolve.ai/blog/Top-5-Benefits](https://resolve.ai/blog/Top-5-Benefits)

[Credential-related source omitted]

格式遵循 [Keep a Changelog](https://keepachangelog.com/)。

帮我安装这个 skill：https://github.com/AlphaMao1/AlphaMao_Skills/tree/main/skills/technology-mapping

<http://xhslink.com/o/9vGoxXSqLrS>

<?xml version="1.0" encoding="UTF-8" standalone="no"?> <!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN"  "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">

<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN"  "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd"> <!-- Generated by graphviz version 14.1.3 (20260303.0454)

"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">

<!-- Title: NeuromorphicComputing_v4 Pages: 1 --> <svg width="3871pt" height="2128pt"  viewBox="0.00 0.00 3871.00 2128.00" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">

<svg width="3871pt" height="2128pt"  viewBox="0.00 0.00 3871.00 2128.00" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink"> <g id="graph0" class="graph" transform="scale(2.08333 2.08333) rotate(0) translate(36 98

viewBox="0.00 0.00 3871.00 2128.00" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">

Executable file

[Credential-related source omitted]

[Credential-related source omitted]

<?xml version="1.0" encoding="UTF-8" standalone="no"?> <!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN"  "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">

<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN"  "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd"> <!-- Generated by graphviz version 14.1.3 (20260303.0454)

"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">

<!-- Title: MCF Pages: 1 --> <svg width="4320pt" height="2650pt"  viewBox="0.00 0.00 4320.00 2650.00" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">

<svg width="4320pt" height="2650pt"  viewBox="0.00 0.00 4320.00 2650.00" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink"> <g id="graph0" class="graph" transform="scale(1.63105 1.63105) rotate(0) translate(36 15

viewBox="0.00 0.00 4320.00 2650.00" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">

Executable file

<?xml version="1.0" encoding="UTF-8" standalone="no"?> <!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN"  "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">

<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN"  "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd"> <!-- Generated by graphviz version 14.1.3 (20260303.0454)

"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">

<!-- Title: BCI_TechMap Pages: 1 --> <svg width="9023pt" height="2360pt"  viewBox="0.00 0.00 9023.00 2360.00" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">

<svg width="9023pt" height="2360pt"  viewBox="0.00 0.00 9023.00 2360.00" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink"> <g id="graph0" class="graph" transform="scale(2.08333 2.08333) rotate(0) translate(36 10

viewBox="0.00 0.00 9023.00 2360.00" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">

Executable file

Executable file

API: https://query.wikidata.org/sparql (free, no auth required)

WIKIDATA_SPARQL_ENDPOINT = "https://query.wikidata.org/sparql" USER_AGENT = "TechMapping-Skill/2.1.0 (technology-mapping; research)" REQUEST_DELAY = 1.0  # seconds between requests (polite rate limiting)

try:             r = requests.get(

r = requests.get(

帮我安装这个 skill：https://github.com/AlphaMao1/AlphaMao_Skills/tree/main/skills/ui-showreel-forge

<http://xhslink.com/o/7LmFimZYx8m>