Learning & Documentation Agent Skills for DeepSeek Harness

Research knowledge bundles and query database schemas.

Transforms books, PDFs, and documentation into actionable Claude Code skills with source-mapped decision rules and playbooks.

Extracts knowledge from Amp threads and updates project docs.

Extract durable knowledge from human input.

Extract tribal knowledge from code, documentation, and commit history to preserve institutional memory

Knowledge base maintenance — lint raw/ and compiled/ for stale, unreferenced, missing-type, and oversized artifacts.

Write and maintain knowledge base articles from resolved support issues.

Turn PDFs, Markdown, text, and web source material into a native-Chinese, visual, interactive, self-contained offline HTML course.

Turn PDFs, Markdown, text, and web source material into a native-English, visual, interactive, self-contained offline HTML course.

Operate a local knowledge-base vault (.vault/ directory) within any project.

A relentless interview that sharpens a plan or design and records the decisions as ADRs and glossary entries.

Write Korean that sounds human. Not formal, not robotic, not AI-generated.

Formats and compiles official Korean public sector proposals, reports, and business plans into compliant Word (.docx) documents with…

Generates legally compliant privacy policies, terms of service, and Next.js consent UI components tailored to Korean PIPA, EU GDPR, and US…

Generates South Korean PIPA-compliant privacy policies, terms of service, consent modals, and cookie banners for Next.js applications.

Translates Korean Markdown prompts into nuanced English versions while maintaining original intent and technical context.

Translate English technical content to Korean using project-specific dictionaries

Generates synchronized, compliant Privacy Policies, Terms of Service, and React legal components for South Korea (PIPA) and Japan (APPI).

Classify curriculum content into Know, Understand, and Do categories to align teaching and assessment approaches.

Use when a draft needs the AI tells removed that vocabulary filters miss, such as em dashes, rhetorical fragments, lists of three, colon…

Slow down interpretation from observation to action.