Topic learn

个人 Agent Skills：健康、阅读、写作、日记、Keep 等脚本与说明

Install

npx -y skills add zhijunio/skills --skill topic-learn

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Learn one topic end to end (学): scoped mission, layered sources, mastery map, verified evidence, canonical article, interview layer, learning records. Modes new or refresh. Use for 主题学习, 技术深挖, 溯源验证; not single-link summaries, blank-page article writing, or codebase audit.

SKILL.md

4.2 KB, ~1.0k tokens by cl100k_base, as published. Nobody here has run it

Topic Learn

Role: 学 — Turn one knowledge point into a reusable, source-backed, verifiable topic asset.

Examples: Java reflection, TCP congestion control, Raft, HTTP caching, Postgres MVCC.

Not for: single-link summaries, opinion without sources, shallow cheat sheets, marketing-only article drafts, codebase audit.

Invocation

Required	Optional
Topic (one knowledge point)	`mode`: `new` (default) or `refresh`
Boundary + out of scope	Depth: `brief` / `standard` (default) / `deep`
Success looks like (≥2 outcomes)	Save path, language `zh` / `en`
Save directory

Missing Success or boundary → ask at most 3 questions. Do not start Source until Mission is writable.

Depth

Level	Deliverables
`brief`	Mission + map + `sources/INDEX` + slim main article (problem, mechanism, sources); interview: one-liner + ≥3 follow-ups
`standard`	Full checklist; glossary optional
`deep`	+ glossary and/or `reference/`; 0–2 side articles; mechanism-heavy topics default to diagrams (`article-structure.md`)

Detail: references/formats.md → Depth.

Modes (`TOPIC.yaml` → `mode`)

Mode	When
`new`	No workspace yet, or full first pass
`refresh`	Merge, gap-fill, re-verify, or humanize-only

Refresh variants (detail in references/workflow.md):

Merge — align scattered drafts; diff-only Write
Gap-fill — only missing pieces (e.g. interview)
Verify / polish — re-run demos, fix verified_on, humanize touched files

Before you start

Read existing topic dir; set mode in TOPIC.yaml (references/formats.md).
New workspace — optional bootstrap:

/path/to/topic-learn/scripts/topic_scaffold.sh /path/to/<topic-dir> <slug> [zh|en]

按需读 — start at references/README.md; never bulk-read all references.

Flow

Scope → Source → Model → Verify → Write → Interviewize → Record → Humanize → Checklist

new: full flow (references/workflow.md).
refresh: delta on each step; touch ≤6 files unless full rewrite.

Stop points (user may say「一次跑完」):

After Scope — confirm Mission + boundary.
After Model — confirm mastery-map + gaps before Write.

Hard rules

One topic per run.
No mastery map → no Write.
No L1/L2 for key claims → no Write (or explicit Open).
No verification → not complete.
No interview layer → not complete.
All prose artifacts in one topic directory (code_repo for runnable code when documented).

Humanize boundary

Polish main + interview when the user asks — preserve sources, verification commands, claims, and glossary terms. User style notes override default tone.

Need	Skill
Learning-topic polish in this workspace	This skill (Step 8)
Neutral de-AI only, no author persona	`humanizer` (user triggers separately)
Author voice + publish-ready in writing flow	`article-write` refine stage

Related skills

Skill	When
`article-read`	Source step — fetch URL/PDF into `sources/` (`references/source-ingest.md`)
`humanizer`	Finished main/interview, neutral de-AI only
`article-write`	Turn mastery assets into publish-oriented articles (user triggers separately)
`flowguard`	Long `new`/`refresh` spanning many sessions — wrap with checkpoints

Validation

bash -n topic-learn/scripts/topic_scaffold.sh
bash -n topic-learn/scripts/topic_check.sh
topic-learn/scripts/topic_scaffold.sh /tmp/topic-smoke test-slug zh
topic-learn/scripts/topic_check.sh /tmp/topic-smoke    # expect warnings on fresh scaffold

Before marking phase: shipped, run topic_check.sh on the topic directory.

References

Start at references/README.md. Scripts: topic_scaffold.sh, topic_check.sh.

What ships with it: 12 files

25.5 KB alongside SKILL.md, 2 of them executable

evals/

topic-learn-evals.json2.1 KB

references/

scripts/

topic_check.shruns4.4 KB
topic_scaffold.shruns2.2 KB

templates/

Gives 0 of the 12 instructions most learn study skills give in ~1.0k tokens

Counted across 546 of the 573 authors here whose files we hold, read 2026-08-07

Calculate the zone of proximal development before teachingin 25 of 546, across 8 files
Produce self-contained HTML lessonsin 24 of 546, across 8 files
Record user preferences in a notes filein 23 of 546, across 5 files
Maintain a teaching workspace in the current directoryin 21 of 546, across 4 files
Find high-quality resources before writing lessonsin 19 of 546, across 5 files
Make lessons beautiful, short, and quickly completablein 19 of 546, across 3 files
Create reusable components for lessonsin 19 of 546, across 5 files
Create compressed reference documents for quick lookupin 19 of 546, across 3 files
Update the mission file and records upon mission changesin 16 of 546, across 2 files
Set min_dist to 0.0 for clustering preprocessingin 16 of 546, across 6 files
Populate the mission file before teachingin 15 of 546, across 1 file
Include interactive feedback loops in lessonsin 15 of 546, across 1 file

Said here and by no other author read

process one topic per run
ask at most three questions if scope is missing
do not start sources until mission is writable
read existing topic directory before starting
set mode in TOPIC.yaml
read references on demand

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.