AI Agent Hub
Back to skills
📚

Chengeng KB Curation L0-L5

Knowledge Management Updated 2026.08.30

Paste the following prompt into your AI chat to install this skill:

Please follow https://skillhub.cn/install/skillhub.md to install @user_88e00a20/chengeng-kb-l0-l5.

About this skill

Problem

If you have project docs, course decks, asset folders, question-bank exports, and scattered PDFs, the hard part is often not “making better summaries” but turning them into a retrievable, traceable, reusable knowledge system. This skill is scoped to a knowledge-base case, not a single PDF or meeting-notes summary. It helps teams assess where the current corpus sits across L0-L5 and then add directories, source audits, object schemas, retrieval tests, and acceptance checks.

How It Works

  • Layer judgment before action: the default output starts with input validation, a one-sentence verdict, knowledge-base type, L0-L5 maturity assessment, standard directory suggestions, database and object schema recommendations, source-priority strategy, retrieval test plan, acceptance checklist, and next steps.
  • Source traceability first: each knowledge object should retain at least source_ref, document_id, and library_path, so answers can be traced back to the source document.
  • Treat intermediate artifacts carefully: PDFs, images, audio, and video are treated as asset registration before OCR or transcript caches exist; the model should work from evidence_packet, object summaries, and sampled original text rather than raw unregistered files.
  • L5 as the completion bar: a version is complete only after unified query, source traceability, source priority, synonym handling, idempotence, and cost logging have been validated.

Boundaries

  • Not for summarizing a single PDF, recording, screenshot, or Markdown file.
  • Not for producing only a summary, reading note, or meeting minutes.
  • It does not promise that all materials are already usable, nor does it treat low-confidence OCR or poor-quality transcripts as reliable knowledge objects.
  • It should not output API keys, tokens, or account credentials, and it should not invent database tables, object counts, or retrieval results.

Use Cases

  • Build a project-doc, course-deck, and question-bank knowledge base and assess its current L0-L5 layer.
  • Design object schemas, directory standards, and source priorities so each object can trace back to `source_ref`.
  • Create retrieval tests and an acceptance checklist to verify unified query, synonyms, idempotence, and traceability.
  • Extract standards, method notes, and asset files into reusable knowledge objects instead of producing one big summary.

Best For

  • Technical owners of knowledge-base projects who need a retrievable, traceable, long-term knowledge base.
  • Content operations managers for courses or training who need to objectify decks, question banks, and assets.
  • Engineers building internal knowledge platforms who need object schemas, source audits, and retrieval acceptance.
  • Knowledge managers maintaining enterprise topic libraries who need L0-L5 assessment and acceptance checklists.