Karpathy LLM Wiki
Astro-Han/karpathy-llm-wiki
A skill your agent uses when building or maintaining a personal LLM-powered knowledge base.
Query and maintain the user's local LLM Wiki from any Codex or Claude Code conversation or project.
Install LLM Wiki
$ npx skills add tudoumashu/ai-memory-skillpack --skill llm-wiki -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install tudoumashu/ai-memory-skillpack llm-wiki --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/tudoumashu/ai-memory-skillpack.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/llm-wiki .claude/skills/llm-wiki && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "llm-wiki" agent skill from https://github.com/tudoumashu/ai-memory-skillpack/tree/main/skills/llm-wiki into .claude/skills/llm-wiki/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "llm-wiki", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/tudoumashu/ai-memory-skillpack/tree/main/skills/llm-wikiType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add tudoumashu/ai-memory-skillpack --skill llm-wiki -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install tudoumashu/ai-memory-skillpack llm-wiki --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/tudoumashu/ai-memory-skillpack.git skills-src && mkdir -p .agents/skills && cp -r skills-src/skills/llm-wiki .agents/skills/llm-wiki && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "llm-wiki" agent skill from https://github.com/tudoumashu/ai-memory-skillpack/tree/main/skills/llm-wiki into .agents/skills/llm-wiki/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "llm-wiki", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add tudoumashu/ai-memory-skillpack --skill llm-wiki -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install tudoumashu/ai-memory-skillpack llm-wiki --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/tudoumashu/ai-memory-skillpack.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/skills/llm-wiki .cursor/skills/llm-wiki && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "llm-wiki" agent skill from https://github.com/tudoumashu/ai-memory-skillpack/tree/main/skills/llm-wiki into .cursor/skills/llm-wiki/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "llm-wiki", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/tudoumashu/ai-memory-skillpack.git --path skills/llm-wiki--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add tudoumashu/ai-memory-skillpack --skill llm-wiki -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install tudoumashu/ai-memory-skillpack llm-wiki --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/tudoumashu/ai-memory-skillpack.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/skills/llm-wiki .gemini/skills/llm-wiki && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "llm-wiki" agent skill from https://github.com/tudoumashu/ai-memory-skillpack/tree/main/skills/llm-wiki into .gemini/skills/llm-wiki/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "llm-wiki", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install tudoumashu/ai-memory-skillpack llm-wikiInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add tudoumashu/ai-memory-skillpack --skill llm-wiki -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/tudoumashu/ai-memory-skillpack.git skills-src && mkdir -p .github/skills && cp -r skills-src/skills/llm-wiki .github/skills/llm-wiki && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "llm-wiki" agent skill from https://github.com/tudoumashu/ai-memory-skillpack/tree/main/skills/llm-wiki into .github/skills/llm-wiki/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "llm-wiki", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add tudoumashu/ai-memory-skillpack --skill llm-wiki -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install tudoumashu/ai-memory-skillpack llm-wiki --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/tudoumashu/ai-memory-skillpack.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/skills/llm-wiki .opencode/skills/llm-wiki && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "llm-wiki" agent skill from https://github.com/tudoumashu/ai-memory-skillpack/tree/main/skills/llm-wiki into .opencode/skills/llm-wiki/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "llm-wiki", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
- Skill name
llm-wiki- GitHub stars
- 456
- Token cost
- ~1.7k tokens
- SKILL.md length
- 723 words
- Files
- 2
- Skills in repo
- 3
- Repo updated
- First seen
- Licence
- MIT
At a glance
Query and maintain the user's local LLM Wiki from any Codex or Claude Code conversation or project.
- Works in 7 steps: Start from 60-Wiki/index.md. → If the query mentions a local project… → If the entity points to a repo docs/ai/,… → …
- The user asks to search personal knowledge
- SKILL.md covers Core Rule, Obsidian Markdown Rules, Query Workflow and Writeback Workflow, plus 3 more sections
- Calls rg and npx
What it does
LLM Wiki is an agent skill from tudoumashu/ai-memory-skillpack. Query and maintain the user's local LLM Wiki from any Codex or Claude Code conversation or project. Use when the user asks to search personal knowledge, recall prior research or decisions, answer from the wiki, create durable notes, ingest a source into the wiki, update topic/concept/entity/question pages, write back useful answers, or run wiki lint/reindex workflows.
Its SKILL.md is about 1.7k tokens, which your agent loads only when the skill is triggered. The skill folder holds 2 other files (for example `agents/openai.yaml`).
It sits in Knowledge Management, covering LLM wikis. The repository describes itself as: Bounded project-memory skills for Codex CLI and Claude Code. The licence is MIT.
When your agent uses it
- The user asks to search personal knowledge
- Recall prior research
- Answer from the wiki
- Create durable notes
Example prompts
- “/llm-wiki”
Requirements
- Node.js
Workflow steps
7 steps, taken from the first numbered list in SKILL.md.
- Start from 60-Wiki/index.md.
- If the query mentions a local project name or repo path, read the central project entity in 60-Wiki/entities/ first. Do not assume the…
- If the entity points to a repo docs/ai/, follow project read discipline — full read only project-card.md and handoff.md; targeted rg for…
- Query the compiled wiki
- If insufficient, widen scope
- Read cited wiki pages directly before making important claims.
- If wiki evidence is stale or insufficient, say so and use source pages, raw sources, project-local docs/ai/, or current web search when…
What it can do on your machine
Read from SKILL.md and the folder at commit f85cdec. It shows what the files ask for, not the result of running them.
Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Shell commands in SKILL.md call:
rgnpx
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md. Its commands use npx, which can reach the network depending on how they are called.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Context cost
LLM Wiki loads about 1.7k tokens when it runs. Until then it costs about 95 tokens; SKILL.md has 723 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
Safety
Auto-check passed
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.
SKILL.md
The full file from tudoumashu/ai-memory-skillpack at commit f85cdec, republished under its MIT licence (© tudoumashu). 723 words, ~1,723 tokens.
Download SKILL.mdSave it as
.claude/skills/llm-wiki/SKILL.md (or your agent's skills folder). This skill also uses 1 other file; get the full folder from GitHub.- name
- llm-wiki
- description
- Query and maintain the user's local LLM Wiki from any Codex or Claude Code conversation or project. Use when the user asks to search personal knowledge, recall prior research or decisions, answer from the wiki, create durable notes, ingest a source into the wiki, update topic/concept/entity/question pages, write back useful answers, or run wiki lint/reindex workflows.
LLM Wiki
Core Rule
Use the local LLM Wiki as the durable knowledge source. Do not copy the wiki into this skill.
Paths on this machine:
VAULT_ROOT=/home/shiyi/Apps/Obsidian/vault
WIKI_ROOT=/home/shiyi/Apps/Obsidian/vault/60-Wiki
INDEX=/home/shiyi/Apps/Obsidian/vault/60-Wiki/index.md
LOG=/home/shiyi/Apps/Obsidian/vault/60-Wiki/log.md
QUERY=/home/shiyi/Apps/Obsidian/vault/70-System/scripts/query_wiki.sh
LINT=/home/shiyi/Apps/Obsidian/vault/70-System/scripts/lint_wiki.sh
REINDEX=/home/shiyi/Apps/Obsidian/vault/70-System/scripts/reindex_qmd.sh
EMBED_SAFE=/home/shiyi/Apps/Obsidian/vault/70-System/scripts/embed_qmd_safe.shDefault wiki prose to Chinese. Keep commands, paths, filenames, config keys, API fields, code, package names, versions, URLs, and fixed English terms unchanged.
Obsidian Markdown Rules
Apply when creating or editing Markdown inside VAULT_ROOT, especially WIKI_ROOT.
- Preserve existing YAML frontmatter. New wiki pages use the local schema keys:
title,created,updated,kind,status,source_paths,tags,summary. - Every agent edit sets frontmatter
updated_by:to the fixed actor stringclaude-code/fable-5orcodex/gpt-5.6-sol-pro, and refreshesupdated. - Do not translate or rewrite frontmatter keys, wikilink targets, block IDs, embed targets, file paths, URLs, code blocks, JSON, or shell commands.
- Use
[[wikilinks]]for internal vault notes ([[path|display]]for Chinese labels); normal Markdown links for external URLs. - Prefer linking related
source/topic/concept/entity/questionpages over leaving isolated notes. - Embeds
![[...]], callouts,%% comments %%, Mermaid, math, footnotes: only when they add durable structure or evidence. Tags live in frontmattertags, not inline. - After editing wiki Markdown, run
lint_wiki.sh.
Query Workflow
- Start from
60-Wiki/index.md. - If the query mentions a local project name or repo path, read the central project entity in
60-Wiki/entities/first. Do not assume the entity file name matches the project name — locate it first withrg -ilF "<project-name-or-repo-path>" /home/shiyi/Apps/Obsidian/vault/60-Wiki/entities/; on multiple hits pick the page whose frontmattersource_pathscontains the repo root (a body mention does not count, same criterion as$ai-project-memory), then read that page. - If the entity points to a repo
docs/ai/, follow project read discipline — full read onlyproject-card.mdandhandoff.md; targetedrgforarchitecture.md,runbook.md,gotchas.md,diagrams/README.md, current ADRs indecisions/; do not readhistory/,reports/,screenshots/,learning/, superseded ADRs (Status: superseded/deprecated), or deprecated ledger files by default(change-log 已停用 2026-07-28,git 提交即账本); on explicit user recall, targeted reads are allowed with cross-checking against current code. - Query the compiled wiki:
/home/shiyi/Apps/Obsidian/vault/70-System/scripts/query_wiki.sh "<query>" --scope wiki- If insufficient, widen scope:
/home/shiyi/Apps/Obsidian/vault/70-System/scripts/query_wiki.sh "<query>"
/home/shiyi/Apps/Obsidian/vault/70-System/scripts/query_wiki.sh "<query>" --scope raw- Read cited wiki pages directly before making important claims.
- If wiki evidence is stale or insufficient, say so and use source pages, raw sources, project-local
docs/ai/, or current web search when required.
Show full SKILL.md (367 more words)Show less
Writeback Workflow
Write back only when the answer or change has durable value. Choose the smallest correct destination:
60-Wiki/questions/: reusable answer.60-Wiki/topics/: cross-source synthesis.60-Wiki/concepts/: reusable method, pattern, term.60-Wiki/entities/: person, company, tool, project, system.60-Wiki/sources/: single-source summary.
When writing back:
- Prefer updating an existing page over creating a duplicate.
- Preserve frontmatter and wikilinks; set
updated_byper the rule above. - Add or update
source_paths; link related pages. - Update
60-Wiki/index.mdwhen navigation changes. log.mdis newest-first: agents log content changes only(ingest、writeback、页面重构)by inserting a## [YYYY-MM-DD] <kind> | <slug>entry above the first existing##heading. Never hand-write query or lint entries;lint_wiki.shitself inserts a dailylint | wiki-healthentry and writes a report under70-System/reports/— tool behavior, not a violation; do not remove those entries.- Keep uncertainty explicit with
inferred,unknown, or open questions. - Never write secret values into any wiki page.
- The vault is a git repository — after content writebacks, commit with an explicit file list including lint by-products(the updated
log.mdand the new70-System/reports/wiki-lint-<date>.md)(git 即账本,不积压未提交改动).
For a user-owned project with repo-local docs/ai/, the central project entity stays lightweight and points to the repo pack: repo path, docs/ai/ path, key file links, short summary, important gaps. Never copy full docs/ai/ content into the wiki; sync only on entity-level changes.
Ingest Workflow
- Read the source.
- Create or update a source page in
60-Wiki/sources/. - Extract durable points; update relevant topic / concept / entity / question pages — do not stop after only a source page when clear updates exist.
- Update
60-Wiki/index.md; add a content entry to60-Wiki/log.mdper Writeback rule 5(newest-first, same header format). - Run lint and reindex.
Validation
/home/shiyi/Apps/Obsidian/vault/70-System/scripts/lint_wiki.sh
/home/shiyi/Apps/Obsidian/vault/70-System/scripts/reindex_qmd.sh llm-wikiLint is required after any wiki edit. Reindex only when future search should find new or changed Markdown. Do not run qmd embed by default; run embedding only for immediate vector-search freshness or batch maintenance:
/home/shiyi/Apps/Obsidian/vault/70-System/scripts/embed_qmd_safe.sh llm-wikiAvoid npx --yes @tobilu/qmd ... embed(version drift, startup cost). Large high-churn files(central log.md, project handoff.md, generated reports)cause repeated expensive embedding — keep entities lightweight and avoid embedding large dumps when a link is enough. Status check: QMD_LLAMA_GPU=false qmd --index llm-wiki status.
Final Response
Mention: wiki pages read; wiki pages changed; validation commands run; any source gaps, uncertainty, or stale information.
© tudoumashu, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Files
SKILL.md and 1 other file in skills/llm-wiki of tudoumashu/ai-memory-skillpack.
- SKILL.md
- agents/openai.yaml
Open the folder on GitHubat commit f85cdec
Compare with similar skills
LLM Wiki next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| LLM Wiki this skilltudoumashu/ai-memory-skillpack | 456 | — | ~1.7k | Automated safety check: Pass | MIT | |
| Karpathy LLM WikiAstro-Han/karpathy-llm-wiki | 2.5k | — | ~3.6k | Automated safety check: Pass | MIT | |
| LLM Wikilewislulu/llm-wiki-skill | 655 | — | ~3.7k | Automated safety check: Pass | None | |
| Wiki Builderrohitg00/pro-workflow | 2.9k | — | ~1k | Automated safety check: Pass | None | |
| Codex History IngestAr9av/obsidian-wiki | 3.5k | — | ~2.2k | Automated safety check: Notes | MIT | |
| Arkon Editnduckmink/arkon | 1.5k | — | ~1.6k | Automated safety check: Pass | Custom licence |
Similar skills
LLM Wiki
lewislulu/llm-wiki-skill
Build and maintain a Karpathy-style LLM knowledge base — a self-compiling Obsidian markdown wiki where an Agent ingests raw sources, compiles cross-linked concept/entity/summary pages, answers…
Wiki Builder
rohitg00/pro-workflow
Start, structure, and grow a persistent research wiki indexed in pro-workflow's SQLite knowledge base.
Codex History Ingest
Ar9av/obsidian-wiki
Ingest Codex CLI conversation/session history into Obsidian as distilled knowledge.
Arkon Edit
nduckmink/arkon
Propose or directly apply edits to Arkon wiki pages, including proposing brand new pages.
Research Wiki Builder
dair-ai/dair-academy-plugins
Creates and maintains configurable research wikis: scaffold a folder, add sources, compile pages and indexes, and file query answers back.
More from tudoumashu/ai-memory-skillpack
Project Mastery Coach
tudoumashu/ai-memory-skillpack
Train strict project ownership from repo-local docs/ai memory and central LLM Wiki project entities.
AI Project Memory
tudoumashu/ai-memory-skillpack
Maintain bounded, durable AI project memory in a repository's docs/ai/ pack.
Related
Categories
Questions about LLM Wiki
What does LLM Wiki do?
Query and maintain the user's local LLM Wiki from any Codex or Claude Code conversation or project. LLM Wiki is an agent skill from tudoumashu/ai-memory-skillpack. Query and maintain the user's local LLM Wiki from any Codex or Claude Code conversation or project.
When should I use LLM Wiki?
LLM Wiki fits situations like: the user asks to search personal knowledge; recall prior research; answer from the wiki; create durable notes.
How do I install LLM Wiki in Claude Code?
Run `npx skills add tudoumashu/ai-memory-skillpack --skill llm-wiki -a claude-code`. Or copy the skill folder (skills/llm-wiki in tudoumashu/ai-memory-skillpack) into .claude/skills/llm-wiki in your project. Claude Code loads it when a task matches its description.
How do I install LLM Wiki in Codex?
Run `npx skills add tudoumashu/ai-memory-skillpack --skill llm-wiki -a codex`. Or copy the skill folder (skills/llm-wiki in tudoumashu/ai-memory-skillpack) into .agents/skills/llm-wiki in your project. Codex loads it when a task matches its description.
Can I use LLM Wiki in Cursor, Gemini CLI or GitHub Copilot?
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add tudoumashu/ai-memory-skillpack --skill llm-wiki -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/llm-wiki, .gemini/skills/llm-wiki, .github/skills/llm-wiki and .opencode/skills/llm-wiki in your project.
What does LLM Wiki need to run?
Going by SKILL.md and its folder, LLM Wiki needs the command-line tools its instructions call (rg and npx). Our summary lists: Node.js.
Does LLM Wiki access the network?
SKILL.md contains no URLs. Its commands use npx, which can reach the network depending on how they are called. This is read from the text; nothing was executed.
Is LLM Wiki safe to install?
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.
What licence does LLM Wiki use?
LLM Wiki is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does LLM Wiki use?
About 1.7k tokens (SKILL.md is roughly 6.9k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.
What are the alternatives to LLM Wiki?
Skills that share tags, products or a category with LLM Wiki: Karpathy LLM Wiki (Astro-Han/karpathy-llm-wiki, 2.5k stars), LLM Wiki (lewislulu/llm-wiki-skill, 655 stars), Wiki Builder (rohitg00/pro-workflow, 2.9k stars) and Codex History Ingest (Ar9av/obsidian-wiki, 3.5k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains LLM Wiki?
tudoumashu (a GitHub user) maintains it in tudoumashu/ai-memory-skillpack, which has 456 GitHub stars. The repository holds 3 skills in this directory. The repository was last updated on September 8, 2026.
Source: tudoumashu/ai-memory-skillpack on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.