By bmentges
Build and maintain personal knowledge bases with Claude. Bootstrap a wiki, ingest sources, query the knowledge, and lint for health — the LLM Wiki pattern as a plugin.
Bootstrap a new LLM Wiki knowledge base in the current directory. Use when the user wants to set up a second brain, research vault, book companion, course notes, business/team knowledge base, or any persistent LLM-maintained wiki. Scaffolds raw/ and wiki/ directories, writes a tailored CLAUDE.md schema, seeds index/log/overview pages, and configures Obsidian defaults. One-shot per vault; subsequent operations use wiki-ingest, wiki-query, and wiki-lint.
Ingest a source document (article, paper, meeting notes, web clipping, transcript, PDF) into an LLM Wiki vault. Use when the user says "ingest this", "add this source", "process this article", "file this", or drops a new file into raw/. Creates a source summary page, updates entity and concept pages, updates wiki/index.md, and appends to wiki/log.md.
Health-check an LLM Wiki vault. Use when the user says "lint the wiki", "health check the wiki", "clean up the wiki", or on a periodic schedule to keep the wiki healthy as it grows. Finds orphan pages, broken wikilinks, stale claims, missing cross-references, contradictions, frontmatter drift, and index staleness. Auto-fixes safe issues, flags the rest for the user.
Answer a question by searching and synthesizing across an LLM Wiki vault. Use when the user asks "what does the wiki say about…", "query the wiki", "ask the wiki", or any substantive question that should be answered from the vault's accumulated knowledge rather than from general training data. Substantive answers can be filed back as analysis pages so explorations compound.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
A Claude Code plugin for building and maintaining persistent, LLM-maintained knowledge bases in Obsidian format. One install, four skills, and any directory becomes a compounding wiki the LLM writes for you.
Based on Andrej Karpathy's LLM Wiki, extended with scaling patterns from agentmemory. Full acknowledgements in origin/CREDITS.md.
Status. Early, not yet on a public Claude Code marketplace. Install by cloning — see Install.
Stop re-deriving knowledge with RAG every time; have the LLM compile it into a structured wiki that compounds over time. You curate sources and ask questions. The LLM does the rest — summarising, cross-referencing, flagging contradictions, keeping everything consistent.
Obsidian is the IDE. The LLM is the programmer. The wiki is the codebase.
For the long version, read origin/llm-wiki.md (the original pattern) and origin/llm-wiki-v2.md (scaling extensions).
Four skills, each triggered by natural language:
| Skill | Say something like… | What it does |
|---|---|---|
wiki-bootstrap | "set up a new knowledge base here" · "build me a second brain" · "start a research vault" | Inspects the directory, interviews you about use case and domain, scaffolds raw/ + wiki/, writes a tailored CLAUDE.md schema, seeds index.md / log.md / overview.md, configures .obsidian/. One-shot per vault. |
wiki-ingest | "ingest this" · "add this article" · "file this paper" · "process this meeting" | Reads one source, writes a summary page in wiki/sources/, updates entity & concept pages, updates the index, appends to the log. Flags contradictions; does not silently overwrite existing claims. |
wiki-query | "what does the wiki say about X?" · "query the wiki" · "compare A and B from the vault" | Searches the wiki, synthesises an answer with wikilink citations, picks the right output format (prose, table, timeline, mermaid). Offers to file substantive answers as wiki/analysis/ pages so explorations compound. |
wiki-lint | "lint the wiki" · "health check" · "clean up the wiki" | Runs seven checks: frontmatter drift, broken wikilinks, orphans, missing cross-references, contradictions, index freshness, overview staleness. Auto-fixes safe issues, flags the rest for you to decide. |
All four skills read the vault's CLAUDE.md at runtime, so the process is plugin-level but the schema is per-vault — each vault's conventions, entity types, and page shapes travel with it.
0.0–1.0 confidence value; skills weigh older/newer claims and surface contradictions rather than overwriting.wiki/log.md is grep-parseable: grep "^## \[" wiki/log.md gives you the full timeline.raw/; lint never resolves contradictions unilaterally; nothing commits to git unless you ask.Clone the repo once, then register it as a local marketplace in Claude Code.
# 1. Clone anywhere on your machine
git clone https://github.com/bmentges/knowledge-base-template ~/code/llm-wiki
cd ~/code/llm-wiki
Then, inside Claude Code:
/plugin marketplace add ~/code/llm-wiki
/plugin install llm-wiki@llm-wiki
Verify the skills are loaded — start a fresh Claude Code session and run:
/plugin
You should see llm-wiki listed as installed, with four skills: wiki-bootstrap, wiki-ingest, wiki-query, wiki-lint.
cd ~/code/llm-wiki
git pull
Then in Claude Code: /plugin marketplace update llm-wiki.
/plugin uninstall llm-wiki@llm-wiki
/plugin marketplace remove llm-wiki
The skills are plain markdown files and work as user-level skills too. Symlink them in:
mkdir -p ~/.claude/skills
for s in wiki-bootstrap wiki-ingest wiki-query wiki-lint; do
ln -s ~/code/llm-wiki/skills/$s ~/.claude/skills/$s
done
Skills are then available in every Claude Code session, no plugin commands needed.
npx claudepluginhub bmentges/knowledge-base-template --plugin llm-wikiBuild and maintain an LLM-curated personal knowledge base in your project — Andrej Karpathy's LLM Wiki pattern, designed to scale to thousands of pages without becoming a context bottleneck. Now with an optional compiled graph layer for typed, provenance-backed relationships.
Claude + Obsidian knowledge companion. Sets up a persistent, compounding wiki vault (Karpathy's LLM Wiki pattern). v1.7 "Compound Vault" + v1.8 methodology modes close 5 of 5 priority gaps from the May 2026 compass artifact. Ships: substrate alignment with kepano/obsidian-skills, default Obsidian CLI transport, hybrid retrieval (contextual prefix + BM25 + cosine rerank per Anthropic's Sept 2024 research), per-file advisory locking for multi-writer safety, pre-commit verifier agent, AND methodology modes (LYT / PARA / Zettelkasten / Generic) for first-class organizational support no other Claude+Obsidian competitor offers. v1.7.x audit closure: every BLOCKER + HIGH + MEDIUM + LOW finding from the v1.7.0 audit is CLOSED or DEFERRED-with-rationale. Optional DragonScale Memory extension (log folds, deterministic addresses, semantic tiling lint, boundary-first autoresearch).
Complete AI coding workflow system. Self-correcting memory + persistent FTS5-indexed research wikis + auto-research loop + multi-LLM council on a single SQLite store. 33 skills, 8 agents, 22 commands, 37 hook scripts across 24 events. Cross-agent via SkillKit.
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.
Connect to Atlassian products including Jira and Confluence. Search and create issues, access documentation, manage sprints, and integrate your development workflow with Atlassian's collaboration tools.
AI-powered wiki generator for code repositories. Generates comprehensive, Mermaid-rich documentation with dark-mode VitePress sites, onboarding guides, deep research, and source citations. Inspired by OpenDeepWiki and deepwiki-open.