Skip to content

README.md

Latest commit

 

History

History
38 lines (31 loc) · 2.85 KB

README.md

File metadata and controls

38 lines (31 loc) · 2.85 KB
Raw

topics — Topics Index

Languages: English · 日本語

Each topic is a specialized application of the learning methods defined in METHOD.md to a specific technology domain. Structure is consistent across all topics: README.md / playbook/ (specialized prompts 00–09 + a 99-corpus-audit exception) / corpus/ (1 file = 1 source) / sources/ (URL manifest) / coverage-map.md / review-log.md (study state: confidence, grades, due dates).

Topic Status Overview
android Worked example — corpus not bundled (rebuild locally) Android 2026 targeting API 37 / Jetpack Compose / KMP / architecture / build / on-device AI. Ships with a specialized playbook, coverage map, and source list so you can see what a fully built-out topic looks like and rebuild its corpus.

android is the only bundled topic, included as a worked example. Add your own topics with the steps below.

Adding a Topic

Automated via Skill: When you instruct Claude Code to "add a topic", the add-topic Skill (.claude/skills/add-topic/) automatically runs scaffolding → corpus collection → validation → index updates. The following is the equivalent manual procedure (if you want to understand the mechanics / do it by hand).

  1. Run python3 tools/scaffold_topic.py --slug <slug> --topic "<Topic>" --role "<persona>" --context-key <key>_context --lang <en|ja> from the repo root — it creates the tree, copies the 11 playbook files for your language (00–09 + the 99-corpus-audit exception) with {TOPIC}/{ROLE} substituted, and writes the corpus-config.json / coverage-map.md / review-log.md / url-manifest.md skeletons (or do the same by hand: create the dirs and copy your language's playbook-template/ files to plain NN-*.md names)
  2. Pick your domain archetype (METHOD.md §4.5) — it decides where to hunt for sources and which 03-tensions (proxy/real) and 04-trajectory (deprecation/interpretations/blueprint) variants to keep. Delete the unused variants.
  3. Editorial pass: fill the domain variants in 02, the tension seeds in 03, the scope lines in 04/05/99, and add a Confusable-concepts table to coverage-map.md (see topics/android/ for how deep this goes)
  4. Collect the corpus following METHOD.md §4 (four source classes: SNAP via fetch_snapshot.py or rebuild_corpus.py, URL, TRANSCRIPT via fetch_transcript.py, NOTE for paywalled works; WebFetch forbidden, 1 file = 1 source, no merging)
  5. Run python3 tools/validate_corpus.py topics/<new-topic> for Lv1 validation (--check-urls for liveness, --emit-coverage to generate the coverage rows)
  6. Write your retention-goal date into review-log.md, paste playbook/01-chat-config.md into Configure Chat, and update this index table