maintain-ai-learning-wiki

name

maintain-ai-learning-wiki

description

Turn scattered AI learning materials (Markdown, text, DOCX, or PDF) into an incrementally maintained, source-traceable learning wiki with bidirectional links, daily update intake, learning cards, quizzes, review queues, weekly summaries, and website-ready data. Use when organizing an AI knowledge folder, refreshing an existing learning knowledge base, creating a reusable knowledge-maintenance workflow, or preparing the data behind a learning site.

Maintain AI Learning Wiki

Run the workflow

1. Define one learning outcome before processing material, such as “explain an AI Agent workflow and make a small working plan in seven days.” Keep the audience, end result, and review cadence explicit.
2. Copy assets/wiki-config.example.json to the project and set the source directory, output directory, learning goal, and trusted daily-search sources.
3. Read references/data-schema.md before integrating output with a site. Read references/schedule-templates.md before creating a scheduled task.
4. Run the maintainer against the source folder. Re-run the same command after any material changes; it compares content hashes, preserves manual notes outside generated blocks, and does not duplicate pages or relationships.

python3 scripts/maintain_wiki.py \
  --source ./materials \
  --output ./wiki \
  --config ./wiki-config.json

5. Inspect manifest.json and graph.json. Treat each source page as traceable: retain source path, source URL when available, first-seen/update timestamps, and a change record. Correct the original page when evidence changes; never silently append a conflicting replacement.
6. Use the generated bidirectional links as suggestions. Confirm semantic links for important topics, especially claims that guide a learner’s next action. Preserve source citations on both the original page and any update page.
7. Search the configured trusted sources every day. Save normalized findings as JSON and ingest them with --search-results; distinguish new facts, corrections, and unsupported claims. Run the maintainer again to synchronize the graph, cards, quiz prompts, review queue, weekly summary, and site-data.json.

python3 scripts/maintain_wiki.py \
  --source ./materials --output ./wiki --config ./wiki-config.json \
  --search-results ./daily-search-results.json

8. Ask the learner one short retrieval question before showing an explanation. Record incorrect or uncertain answers in the site’s review state, schedule them sooner, and make the next card explain the missing concept with a concrete example. Do not present a static wiki as adaptive learning.
9. Publish only the website-ready fields in site-data.json: cards, quizzes, knowledge-tree branches, sources, update markers, progress, and weekly result. Render updates visibly so a changed conclusion is distinguishable from a new note.
10. Validate before handoff. Confirm source counts, page uniqueness, reciprocal edges, no missing source paths, no duplicated update IDs, valid JSON, and a visible weekly learning outcome.

python3 scripts/maintain_wiki.py --source ./materials --output ./wiki --config ./wiki-config.json --validate

Use the generated artifacts

• Read references/data-schema.md for the contract and required UI mappings.
• Read references/schedule-templates.md for daily and weekly task prompts.
• Use scripts/maintain_wiki.py --help for the complete command contract. Install no dependency for Markdown, text, or DOCX extraction; PDF extraction degrades gracefully if pypdf/PyPDF2 is unavailable.