Menu

#446 by GitHub stars

Astro-Han/karpathy-llm-wiki

Use when building or maintaining a personal LLM-powered knowledge base. Triggers: ingesting sources into a wiki, querying wiki knowledge, linting wiki quality, 'add to wiki', 'what do I know about', or any mention of 'LLM wiki' or 'Karpathy wiki'.

Engineering

History

Review weekly changes to SKILL.md before updating. Dates mark the weeks in which we observed each version; checks apply to the recorded commits.

View weekly changes

SKILL.md

Week of Oct 26, 2026 · +0 / −0 lines

Before Network or broad shell → After Network or broad shell

eafcc77001e4 → eafcc77001e4

Compare on GitHub

View on GitHubAdd to my skills

Verified commit eafcc77001e496cc43499e4923b663aec722c813

2.7kGitHub stars
47stars in 7 days
1skill in this repository
0commits in 30 days

Heat

5.8
  • Star growth, last 7 daysAhead of 90% of tracked repositories
  • Star growth, last 24 hoursAhead of 84% of tracked repositories
  • Rank movement, last 7 daysAhead of 47% of tracked repositories
  • Commits, last 30 daysAhead of 0% of tracked repositories
  • Accounts asking about it through MCP, last 7 daysNot available
  • Hot votes this weekAhead of 0% of tracked repositories

Heat never changes the order by stars, and sponsorship never touches it. How heat works

Stars over time

Aug 2, 2026Nov 6, 20262,691

Individual skills in this repository

How it works

karpathy-llm-wiki

When your agent uses it

Use when building or maintaining a personal LLM-powered knowledge base. Triggers: ingesting sources into a wiki, querying wiki knowledge, linting wiki quality, 'add to wiki', 'what do I know about', or any mention of 'LLM wiki' or 'Karpathy wiki'.

What it loads into context

Frontmatter

73 tokens

Always available to the agent

SKILL.md

3,152 tokens

When the skill is activated

Reference documents

7,020 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

  1. Architecture
  2. The Grounding Invariant
  3. Ingest
  4. Query
  5. Lint
  6. Conventions

Scripts executed separately

  • scripts/check_evidence.py
  • tests/test_check_evidence.py

Files

  • .gitignoreasset38 bytes
  • LICENSEasset1,066 bytes
  • README.mddoc7,852 bytes
  • SKILL.mdcore14,342 bytes
  • SKILL.mdcore14,342 bytes
  • SKILL.mdcore14,342 bytes
  • SKILL.mdcore14,342 bytes
  • SKILL.mdcore14,342 bytes
  • SKILL.mdcore14,342 bytes
  • SKILL.mdcore14,342 bytes
  • SKILL.mdcore14,342 bytes
  • SKILL.mdcore14,342 bytes
  • SKILL.mdcore14,342 bytes
  • SKILL.mdcore14,342 bytes
  • SKILL.mdcore14,342 bytes
  • SKILL.mdcore14,342 bytes
  • SKILL.mdcore14,342 bytes
  • SKILL.mdcore14,342 bytes
  • SKILL.mdcore14,342 bytes
  • assets/karpathy-tweet.pngasset283,104 bytes
  • examples/2026-03-19-claude-code-statusline-landscape.mddoc9,213 bytes
  • examples/README.mddoc1,433 bytes
  • examples/ai-coding-tools-index.mddoc1,295 bytes
  • examples/claude-code-statusline-landscape.mddoc5,429 bytes
  • examples/log-sample.mddoc948 bytes
  • references/archive-template.mddoc752 bytes
  • references/article-template.mddoc1,283 bytes
  • references/index-template.mddoc541 bytes
  • references/raw-template.mddoc310 bytes
  • scripts/check_evidence.pyruns14,475 bytes
  • tests/test_check_evidence.pyruns28,043 bytes
Read SKILL.md

# Karpathy LLM Wiki

Build and maintain a personal knowledge base using LLMs. You manage two directories: `raw/` (immutable source material) and `wiki/` (compiled knowledge articles). Sources go into raw/, you compile them into wiki articles, and the wiki compounds over time.

Core ideas from Karpathy:
- "The LLM writes and maintains the wiki; the human reads and asks questions."
- "The wiki is a persistent, compounding artifact."

## Architecture

Three layers, all under the user's project root:

**raw/** — Immutable source material. You read, never modify. Organized by topic subdirectories (e.g., `raw/machine-learning/`).
View the complete source on GitHub

Skills included

  • karpathy-llm-wiki

Repository facts

License
MIT
Created
Apr 5, 2026
Last pushed
Jul 23, 2026
Forks
316
Open issues
5
Project activity
Slowing

Rank over the last 12 weeks

Now #446 · best #377 · down 50 in 30 days

  1. #437 on Oct 29
  2. #439 from Oct 30 to Nov 1
  3. #441 on Nov 2
  4. #443 on Nov 3
  5. #444 from Nov 4 to Nov 5
  6. #446 since Nov 6
78 earlier periods
  1. #390 on Aug 2
  2. #385 on Aug 7
  3. #387 on Aug 8
  4. #384 on Aug 9
  5. #393 on Aug 10
  6. #391 on Aug 11
  7. #384 on Aug 12
  8. #386 on Aug 13
  9. #387 on Aug 14
  10. #385 on Aug 15
  11. #391 on Aug 16
  12. #388 on Aug 17
  13. #395 on Aug 18
  14. #391 from Aug 19 to Aug 20
  15. #389 on Aug 21
  16. #386 on Aug 22
  17. #392 on Aug 23
  18. #385 on Aug 24
  19. #387 on Aug 25
  20. #386 on Aug 26
  21. #389 on Aug 27
  22. #390 on Aug 28
  23. #386 on Aug 29
  24. #377 on Aug 30
  25. #388 on Aug 31
  26. #395 on Sep 1
  27. #385 on Sep 2
  28. #393 on Sep 3
  29. #383 on Sep 4
  30. #389 on Sep 5
  31. #384 on Sep 6
  32. #392 on Sep 7
  33. #393 from Sep 8 to Sep 9
  34. #394 on Sep 10
  35. #393 on Sep 11
  36. #394 on Sep 12
  37. #392 on Sep 13
  38. #398 on Sep 14
  39. #400 from Sep 15 to Sep 16
  40. #393 on Sep 17
  41. #396 on Sep 18
  42. #387 on Sep 19
  43. #396 on Sep 20
  44. #392 on Sep 21
  45. #399 on Sep 22
  46. #390 on Sep 23
  47. #387 on Sep 24
  48. #384 on Sep 25
  49. #379 on Sep 26
  50. #386 on Sep 27
  51. #388 on Sep 28
  52. #400 on Sep 29
  53. #397 on Sep 30
  54. #396 on Oct 1
  55. #398 on Oct 2
  56. #386 on Oct 3
  57. #384 on Oct 4
  58. #390 from Oct 5 to Oct 6
  59. #396 on Oct 7
  60. #400 on Oct 8
  61. #411 on Oct 9
  62. #399 on Oct 10
  63. #401 on Oct 11
  64. #409 on Oct 12
  65. #405 on Oct 13
  66. #408 on Oct 14
  67. #410 on Oct 15
  68. #416 on Oct 16
  69. #407 on Oct 17
  70. #411 on Oct 18
  71. #416 on Oct 19
  72. #418 on Oct 20
  73. #420 on Oct 21
  74. #419 on Oct 22
  75. #417 on Oct 23
  76. #411 on Oct 24
  77. #432 on Oct 25
  78. #435 on Oct 28

Before you install

Network or broad shell

Automated signals, not a guarantee

Checked Nov 2, 2026

Not fully analyzed: some files were too long, unreadable or missing. This never lowers the level.

Show 15 notes
  • SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
  • SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
  • SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
  • SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
  • SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
  • SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
  • SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
  • SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
  • SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
  • SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
  • SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
  • SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
  • SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
  • SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
  • SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
Show all 43 signals by file

README.md

  • Network accessline 5
    [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)
  • Network accessline 6
    [![GitHub stars](https://img.shields.io/github/stars/Astro-Han/karpathy-llm-wiki?style=social)](https://github.com/Astro-Han/karpathy-llm-wiki)
  • Network accessline 7
    [![GitHub forks](https://img.shields.io/github/forks/Astro-Han/karpathy-llm-wiki?style=social)](https://github.com/Astro-Han/karpathy-llm-wiki)
  • Network accessline 8
    [![Agent Skills](https://img.shields.io/badge/Agent_Skills-compatible-blue)](https://agentskills.io)
  • Network accessline 9
    [![Install](https://img.shields.io/badge/Install-npx_add--skill-green)](https://github.com/Astro-Han/karpathy-llm-wiki#install)
  • Network accessline 15
    `karpathy-llm-wiki` packages [Karpathy's LLM Wiki idea](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f) into one installable [Agent Skills](https://agentskills.io) skill. Your coding agent ingests sources into `raw/`, com
  • Network accessline 56
    Works with any tool that supports the [Agent Skills](https://agentskills.io) standard.
  • Network accessline 64
    > "Ingest this article: https://example.com/attention-is-all-you-need"
  • Network accessline 100
    This skill follows the [agentskills.io](https://agentskills.io) open standard:
  • Network accessline 142
    Unofficial community implementation of the workflow from [Karpathy's LLM Wiki idea](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f). The value here is the reusable workflow, prompt structure, and battle-tested knowledge-c
  • Network accessline 144
    See also: [lucasastorian/llmwiki](https://github.com/lucasastorian/llmwiki), [atomicmemory/llm-wiki-compiler](https://github.com/atomicmemory/llm-wiki-compiler). We are tracking Google's [Open Knowledge Format](https://github.com/GoogleClou

SKILL.md

  • Network accessline 49
    ### Fetch (raw/)

examples/2026-03-19-claude-code-statusline-landscape.md

  • Network accessline 11
    | Project | Stars | Language | Form | Last Update | Features | |---------|------:|----------|------|-------------|----------| | [ccusage](https://github.com/ryoppippi/ccusage) | 11,693 | TypeScript | CLI + statusline | 03-18 | Usage analysi
  • Network accessline 12
    |---------|------:|----------|------|-------------|----------| | [ccusage](https://github.com/ryoppippi/ccusage) | 11,693 | TypeScript | CLI + statusline | 03-18 | Usage analysis + burn rate | | [claude-hud](https://github.com/jarrodwatts/c
  • Network accessline 13
    | [ccusage](https://github.com/ryoppippi/ccusage) | 11,693 | TypeScript | CLI + statusline | 03-18 | Usage analysis + burn rate |
  • Network accessline 14
    | [claude-hud](https://github.com/jarrodwatts/claude-hud) | 7,038 | JavaScript | statusline | 03-15 | Most features, pioneer |
  • Network accessline 15
    | [Claude-Code-Usage-Monitor](https://github.com/Maciek-roboblog/Claude-Code-Usage-Monitor) | 7,009 | Python | Standalone dashboard | 2025-09 | ML prediction, unmaintained |
  • Network accessline 16
    | [ccstatusline](https://github.com/sirmalloc/ccstatusline) | 5,421 | TypeScript | statusline | 03-16 | Highly customizable + themes |
  • Network accessline 17
    | [CCometixLine](https://github.com/Haleclipse/CCometixLine) | 2,227 | Rust | statusline | 03-14 | High-performance binary |
  • Network accessline 18
    | [claude-powerline](https://github.com/Owloops/claude-powerline) | 931 | TypeScript | statusline | 03-18 | vim powerline style |
  • Network accessline 19
    | [kamranahmedse/claude-statusline](https://github.com/kamranahmedse/claude-statusline) | 726 | Shell | statusline | 03-10 | Minimalist |
  • Network accessline 20
    | [claude-code-statusline](https://github.com/rz1989s/claude-code-statusline) | 397 | Shell | statusline | 03-14 | 4-line enhanced + themes |
  • Network accessline 21
    | [claude-code-usage-bar](https://github.com/leeguooooo/claude-code-usage-bar) | 159 | Python | statusline | 2025-11 | burn rate + depletion prediction |
  • Network accessline 22
    | [claudia-statusline](https://github.com/hagan/claudia-statusline) | 21 | Rust | statusline | 2026-01 | SQLite persistence + cloud sync |
  • Network accessline 23
    | [felipeelias/claude-statusline](https://github.com/felipeelias/claude-statusline) | 2 | Go | statusline | 03-17 | Single binary, minimalist |
  • Network accessline 66
    | Channel | Stars | claude-pace Status | |---------|------:|--------------------| | [awesome-claude-code](https://github.com/hesreallyhim/awesome-claude-code) | 29,014 | Not listed |
  • Network accessline 67
    |---------|------:|--------------------| | [awesome-claude-code](https://github.com/hesreallyhim/awesome-claude-code) | 29,014 | Not listed | | Anthropic official plugin directory | 12,653 | Zero statusline listings |
  • Network accessline 68
    | [awesome-claude-code](https://github.com/hesreallyhim/awesome-claude-code) | 29,014 | Not listed |
  • Network accessline 95
    - **Migration from Node.js to lightweight runtimes**: Rust (CCometixLine, claudia-statusline), Go (felipeelias), Shell (kamranahmedse, rz1989s, claude-pace) - **Zero-dep install as selling point**: Go single binary, Bash script curl one-lin
  • Network accessline 96
    - **Zero-dep install as selling point**: Go single binary, Bash script curl one-line install

examples/claude-code-statusline-landscape.md

  • Network accessline 65
    - Rust: CCometixLine, claudia-statusline - Go: felipeelias (single binary, minimal) - Shell/Bash: kamranahmedse, rz1989s (curl one-line install)
  • Network accessline 67
    - Shell/Bash: kamranahmedse, rz1989s (curl one-line install)

scripts/check_evidence.py

  • Executable scriptline 1
    #!/usr/bin/env python3

tests/test_check_evidence.py

  • Executable scriptline 1
    Executable file
  • Network accessline 11
    > Source: https://example.com/ghostty
  • Network accessline 45
    > Source: https://example.com/numbers
  • Network accessline 98
    > Source: https://example.com/x
  • Network accessline 132
    > Source: https://example.com/numbers
  • Network accessline 149
    > Source: https://example.com/plain
  • Network accessline 350
    > Source: https://example.com/notes
  • Network accessline 620
    article = PLAIN_ARTICLE.replace(             "There were 42 users; the ratio was 3.14; founded in 2026.",             "> The [terminal](https://example.com) should feel invisible to users.\n",
  • Network accessline 621
    "There were 42 users; the ratio was 3.14; founded in 2026.",             "> The [terminal](https://example.com) should feel invisible to users.\n",         )
  • Network accessline 622
    "> The [terminal](https://example.com) should feel invisible to users.\n",

How the signals fit together

Facts from a static read of the files at the pinned commit. They describe what the code does, not why.

Signal confidence: 43 facts

Works with

Agent compatibility, from the files in this repository
AgentStatusEvidenceInstall to
Claude CodeCompatibleStandard SKILL.md SKILL.md.claude/skills/<skill-name>/
CodexCompatibleStandard SKILL.md SKILL.md.agents/skills/<skill-name>/
CursorCompatibleStandard SKILL.md SKILL.md.cursor/skills/<skill-name>/
Gemini CLICompatibleStandard SKILL.md SKILL.md.gemini/skills/<skill-name>/
GitHub CopilotCompatibleStandard SKILL.md SKILL.md.github/skills/<skill-name>/
OpenCodeCompatibleStandard SKILL.md SKILL.md.opencode/skills/<skill-name>/
  • Native The repository ships a manifest, folder or frontmatter field for this agent.
  • Compatible A standard SKILL.md that this agent can read.
  • Unknown No evidence either way.

Derived automatically from paths, manifests and frontmatter. Not tested on each agent.

You might also use

ChangWenC/understanding-ladder

Karpathy's understanding ladder as agent skills: auto-pick text, diagram, interactive HTML, or video storyboard for any explanation. Nothing cut. English + 中文.

121GitHub stars

swarmclawai/swarmvault

The local-first LLM Wiki: open-source knowledge graph builder, RAG knowledge base, and agent memory store. Built on Andrej Karpathy's pattern. An Obsidian alternative for personal…

734GitHub stars

Tubo2333/obsidian-knowledge-brain

AI agent skill that remembers every technical decision & bug fix across sessions — and learns from them. v4.0, MIT. | 跨会话记忆的AI编程助手知识大脑

101GitHub stars

ComposioHQ/awesome-claude-skills

A curated list of awesome Claude Skills, resources, and tools for customizing Claude AI workflows

78.9kGitHub stars

inkeep/open-knowledge

Beautiful, AI-native markdown IDE and LLM wiki

4.8kGitHub stars

TbusOS/whetstone

Distill dev sessions into portable Agent Skills — then hold them to an evidence standard you can run: mechanical confidence levels, a >=2-platform promotion gate, no silent overwr…

1GitHub stars

Add this badge to your README

It updates every night and links to this page.

Astro-Han/karpathy-llm-wiki: #446 on hotskills, 2.7k stars

[![Astro-Han/karpathy-llm-wiki on hotskills](https://hotskills.dev/badge/astro-han-karpathy-llm-wiki.svg)](https://hotskills.dev/skill/astro-han-karpathy-llm-wiki/)