# sdyckjq-lab/llm-wiki-skill

基于 Karpathy llm-wiki 方法论的个人知识库构建 Skill，支持多平台！

Repository: https://github.com/sdyckjq-lab/llm-wiki-skill
Canonical: https://ross.abutalabs.com/products/llm-wiki-skill
Language: TypeScript
License Family: other
Last push: 2026-07-27T03:30:44+00:00

## Health v2 (maintenance only)
Score: 57/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 94, release rhythm 35, longevity 10
- inputs: {"age_days": 150, "days_push": 37, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, young, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2378, forks 285 (observed 2026-08-28T04:06:42.178995+00:00)

## What it is
A personal knowledge base Skill based on Andrej Karpathy's llm-wiki methodology, installable into AI CLI agents like Claude Code, Codex, OpenClaw, and Hermes. It compiles fragmented information into a persistent, bidirectionally-linked wiki with an offline interactive knowledge graph, and includes a developing local agent workbench sharing the same graph engine.

## Use cases
- build a personal knowledge base from scattered notes and links
- have my AI agent digest an article into wiki pages
- visualize my notes as an interactive knowledge graph
- maintain a wiki with bidirectional links between entity pages
- track which extracted facts are verified vs inferred
- crystallize valuable chat conversations into knowledge pages
- auto-inject my knowledge base context into agent sessions

## When to choose
- you use Claude Code, Codex, or similar AI CLI agents and want a persistent, self-maintaining knowledge base
- you want offline, self-contained HTML knowledge graphs with no server dependency
- you prefer knowledge compiled once and maintained rather than re-derived per query
- you need confidence annotations (EXTRACTED/INFERRED/AMBIGUOUS/UNVERIFIED) on generated content

## When to avoid
- you need a hosted or multi-user collaborative wiki like Notion or Obsidian Publish
- you want a fully polished desktop app today (the workbench is still in development)
- you are uncomfortable with prompts and retrieved snippets being sent to your model provider
- you require a formally licensed release for compliance (no license file is declared in metadata)

## Facets
- artifact type: plugin
- maturity: active
- function: documentation, markdown, rag, data-visualization, developer-tools
- domain: large-language-models, documentation, developer-tools
- platform: cli, cross-platform
- tags: llm-skill, personal-wiki, knowledge-graph, claude-code, codex, agent-workbench, bidi-links, offline-html, karpathy-llm-wiki, note-taking, knowledge-management, ai-agents, web-server

## Member repositories
- sdyckjq-lab/llm-wiki-skill (main) score 57

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:42.178995+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:34:54.603672+00:00, confidence not recorded.
  - readme: https://github.com/sdyckjq-lab/llm-wiki-skill (fetched 2026-08-28T04:06:42.178995+00:00, sha b5ff2a1ded58)
- Data as of 2026-08-30T08:39:29.467469+00:00.