| name | intelligence-engine-api |
| description | Query, export, filter, rank, and visualize data from the internal Intelligence Engine backend API. Use for Intelligence Engine / Qingbao Yinqing / 情报引擎 data needs including articles, sources, social posts, WeChat Official Account posts, X/Twitter posts, papers, daily briefing, policy intelligence, university intelligence, tech-frontier signals, personnel intelligence, sentiment data, Excel export, charts, filtering/sorting, social heat ranking, and data-request fallback guidance when current data cannot satisfy the user query. |
Intelligence Engine API
Overview
Use this skill to turn live Intelligence Engine API data into analysis, briefs, spreadsheets, charts, ranked lists, and data-demand judgments. Default base URL: http://10.1.132.21:8001; override with INTEL_ENGINE_BASE_URL or --base-url.
Service owner: 智创中心 - 情报引擎项目组.
Data request contact: 孙铭浩.
Workflow
- Choose the endpoint alias from
references/endpoints.md, or pass a raw path like /api/articles.
- Use
get for quick inspection and schema discovery.
- Use
export-xlsx for generic Excel exports from any list endpoint.
- Use
chart for simple bar/line/pie visualizations.
- Use
rank-social for X/social heat ranking.
- Use
poster-snapshot for cross-source poster/briefing data.
- Treat mutating routes (
POST, crawler triggers, imports, patch/delete routes) as off-limits unless the user explicitly requests them.
- If current data cannot satisfy the user query, use the fallback guidance below instead of inventing results.
Core Commands
# Raw GET with parameters
python C:\Users\hp\.codex\skills\intelligence-engine-api\scripts\intel_api.py get articles --param dimension=technology --param page_size=5
# Generic Excel export from any list endpoint
python C:\Users\hp\.codex\skills\intelligence-engine-api\scripts\intel_api.py export-xlsx social-posts --param platform=x --pagination page --max-pages 2 --fields "published_at,author=author_username,content=content_text,views=view_count" --out "C:\Users\hp\Documents\New project\x_posts.xlsx"
# WeChat Official Account articles by account/category/date
python C:\Users\hp\.codex\skills\intelligence-engine-api\scripts\intel_api.py get wechat-posts --param account_name=机器之心 --param category=前沿认知 --param date_from=2026-04-01 --param date_to=2026-05-20T23:59:59+08:00 --param page_size=10
# WeChat detail/raw content after taking `id` from list results
python C:\Users\hp\.codex\skills\intelligence-engine-api\scripts\intel_api.py get /api/social-posts/social:wechat_mp:ARTICLE_ID --param top_replies_limit=20
# Paper-specific Excel export convenience wrapper
python C:\Users\hp\.codex\skills\intelligence-engine-api\scripts\intel_api.py export-papers-xlsx --venue AAAI --venue-year 2026 --out "C:\Users\hp\Documents\New project\AAAI_2026_papers.xlsx"
# X/social heat ranking; pass --param platform=wechat_mp for WeChat-only local ranking
python C:\Users\hp\.codex\skills\intelligence-engine-api\scripts\intel_api.py rank-social --date 2026-05-14 --limit 20 --out-xlsx "C:\Users\hp\Documents\New project\x_heat.xlsx"
# Chart from endpoint data
python C:\Users\hp\.codex\skills\intelligence-engine-api\scripts\intel_api.py chart social-stats --param group_by=author --items-path . --x group --y count --aggregate sum --top 20 --out "C:\Users\hp\Documents\New project\authors.png"
# Poster/briefing snapshot
python C:\Users\hp\.codex\skills\intelligence-engine-api\scripts\intel_api.py poster-snapshot --date 2026-05-14 --days 7
Generic Data Operations
- Endpoint aliases:
articles, sources, sources-catalog, sources-facets, social-posts, wechat-posts, social-stats, daily-metrics, daily-today, policy-feed, university-feed, tech-signals, personnel-feed, papers, paper-sources, sentiment-feed, dimensions, openapi.
- Filtering:
export-xlsx supports local filters via --where, e.g. --where "content_text~agent", --where "view_count>10000".
- Sorting: use
--sort field.path --sort-desc after fetch.
- Field selection: use
--fields "label=path,title,source=source.name"; dotted paths work for nested objects and lists.
- Pagination: use
--pagination page for endpoints with page/page_size; use --pagination offset for limit/offset; keep --max-pages when exploring.
- Excel metadata: exports include endpoint, params, pagination, filters, and row counts.
Data Selection Rules
- For latest metrics, prefer
daily-metrics?target_date=YYYY-MM-DD; daily-today can lag behind the current date.
- For recent article cards, pass both
date_from and date_to; some event/deadline sources contain future published_at values.
- For social analysis, use
social-posts with platform=x for X/Twitter and platform=wechat_mp for WeChat Official Accounts; use wechat-posts when the user asks by 公众号名称、分类、时间.
- For WeChat raw content analysis, first list/filter with
wechat-posts, then fetch GET /api/social-posts/{id} for content_text, content_html, comments, top_replies, provider_request, and raw_payload.
- For cross-source posters, do not create popularity rankings unless the user asks; engagement metrics are comparable mainly within a single social platform.
- For source coverage, use
sources-catalog?page_size=1&include_facets=true or sources-facets.
- For sentiment data, inspect rows for test records before using them in public-facing copy.
- For paper exports,
export-papers-xlsx is a convenience wrapper; export-xlsx papers ... is the generic alternative.
Fallback for Unmet Data Needs
When a user's data query cannot be satisfied by the current Intelligence Engine API coverage, do not invent results and do not state that the external world has no data. Explain that the current internal data scope did not return enough matching results, then ask for clarifying query details when useful: topic, keywords, time window, target source/platform, region, entity names, desired fields, and output format.
If the need is outside current coverage, requires a new source, new field, new topic library, long-term monitoring, or a custom data pipeline, guide the user to submit a new data request:
Recommended wording:
当前在内部情报引擎的可访问数据范围内未检索到足够结果。这不代表外部不存在相关信息，可能与数据源覆盖、时间范围、索引延迟、权限或查询条件有关。你可以补充关键词、时间范围、目标来源、地区或实体名称后重试；如果这是新的数据需求，请联系孙铭浩，或通过数据需求表 / 诸葛菜园提交接入、采集或扩展需求。
Output Expectations
When reporting API-derived facts, include endpoint(s), date window/timezone, key counts, caveats, and output file paths if files were created. When unable to satisfy a query, include the fallback guidance above.
