# Llm Wiki > LLM-powered knowledge base from your Claude Code, Codex CLI, Copilot, Cursor & Gemini sessions. Karpathy's LLM Wiki pattern β€” implemented and shipped. - **Type:** MCP server - **Install:** `agentstack add mcp-pratiyush-llm-wiki` - **Verified:** Yes β€” security-reviewed for prompt injection and unsafe behavior - **Seller:** [Pratiyush](https://agentstack.voostack.com/s/pratiyush) - **Installs:** 0 - **Category:** [AI & ML](https://agentstack.voostack.com/c/ai-and-ml) - **Latest version:** 0.1.0 - **License:** MIT - **Upstream author:** [Pratiyush](https://github.com/Pratiyush) - **Source:** https://github.com/Pratiyush/llm-wiki ## Install ```sh agentstack add mcp-pratiyush-llm-wiki ``` Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent. ## About # llmwiki > **LLM-powered knowledge base from your Claude Code, Codex CLI, Cursor, Gemini CLI, and Obsidian sessions.** > Built on [Andrej Karpathy's LLM Wiki pattern](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f). ## πŸ‘‰ Live demo: **[pratiyush.github.io/llm-wiki](https://pratiyush.github.io/llm-wiki/)** Rebuilt on every `master` push from the synthetic sessions in [`examples/demo-sessions/`](examples/demo-sessions). No personal data. Shows every feature of the real tool (activity heatmap, tool charts, token usage, model info cards, vs-comparisons, project topics) running against safe reference data. [](LICENSE) [](https://www.python.org/) [](CHANGELOG.md) [](tests/) [](https://github.com/Pratiyush/llm-wiki/actions/workflows/ci.yml) [](https://github.com/Pratiyush/llm-wiki/actions/workflows/link-check.yml) [](https://github.com/Pratiyush/llm-wiki/actions/workflows/wiki-checks.yml) [](https://github.com/Pratiyush/llm-wiki/pkgs/container/llm-wiki) [](https://claude.com/claude-code) [](https://github.com/openai/codex) [](https://github.com/features/copilot) [](https://cursor.com) [](https://ai.google.dev/gemini-api) [](https://obsidian.md) --- Every Claude Code, Codex CLI, Copilot, Cursor, and Gemini CLI session writes a full transcript to disk. You already have hundreds of them and never look at them again. **llmwiki** turns that dormant history into a beautiful, searchable, interlinked knowledge base β€” locally, in two commands. Plus, it produces AI-consumable exports (`llms.txt`, `llms-full.txt`, JSON-LD graph, per-page `.txt` + `.json` siblings) so other AI agents can query your wiki directly. ```bash ./setup.sh # one-time install ./build.sh && ./serve.sh # build + serve at http://127.0.0.1:8765 ``` **Want LLM-enriched wiki pages from the CLI?** After `./setup.sh`, `llmwiki sync` only fills `raw/`. To populate `wiki/sources/` (semantic summaries), set `synthesis.backend` in `examples/sessions_config.json` (`ollama` for local Ollama, or keep `dummy` for stubs), then run `llmwiki synthesize` or chain it with `llmwiki all --with-synth`. See [docs/modes/api/](docs/modes/api/) and `llmwiki sync --status` for the current backend hint. ```bash llmwiki sync llmwiki all --with-synth --graph-engine builtin # synthesize β†’ build β†’ graph β†’ export β†’ lint ``` **Contributing in one line:** read [`CONTRIBUTING.md`](CONTRIBUTING.md), keep PRs focused (one concern each), use `feat:` / `fix:` / `docs:` / `chore:` / `test:` commit prefixes, never commit real session data (`raw/` is gitignored), no new runtime deps. CI must be green to merge. ## Screenshots All screenshots below are from the **public demo site** which is built on every `master` push from the dummy example sessions. Your own wiki will look identical β€” just with your real work. ### Home β€” projects overview with activity heatmap ### All sessions β€” filterable table across every project ### Session detail β€” full conversation + tool calls ### Changelog β€” renders `CHANGELOG.md` as a first-class page ### Projects index β€” freshness badges + per-project stats ## What you get ### Human-readable - **All your sessions**, converted from `.jsonl` to clean, redacted markdown - **A Karpathy-style wiki** β€” `sources/`, `entities/`, `concepts/`, `syntheses/`, `comparisons/`, `questions/` linked with `[[wikilinks]]` - **A beautiful static site** you can browse locally or deploy to GitHub Pages - Global search (Cmd+K command palette with fuzzy match over pre-built index) - [highlight.js](https://highlightjs.org/) client-side syntax highlighting (light + dark themes) - Dark mode (system-aware + manual toggle with `data-theme`) - Keyboard shortcuts: `/` search Β· `g h/p/s` nav Β· `j/k` rows Β· `?` help - Collapsible tool-result sections (auto-expand > 500 chars) - Copy-as-markdown + copy-code buttons - Breadcrumbs + reading progress bar - Filter bar on sessions table (project/model/date/text) - Reading time estimates (`X min read`) - Related pages panel at the bottom of every session - Activity heatmap on the home page - Model info cards with structured schema (provider, pricing, benchmarks) - Auto-generated vs-comparison pages between AI models - Append-only changelog timeline with pricing sparkline - Project topic chips (GitHub-style tags on project cards) - Agent labels (colored badges: Claude/Codex/Copilot/Cursor/Gemini) - Recently-updated card on the home page - Dataview-style structured queries in the command palette - Hover-to-preview wikilinks - Deep-link icons next to every heading - Mobile-responsive + print-friendly ### AI-consumable (v0.4) Every HTML page has sibling machine-readable files at the same URL: - `.html` β€” human HTML with schema.org microdata - `.txt` β€” plain text version (no HTML tags) - `.json` β€” structured metadata + body Site-level AI-agent entry points: | File | What | |---|---| | [`/llms.txt`](https://llmstxt.org) | Short index per [llmstxt.org spec](https://llmstxt.org) | | `/llms-full.txt` | Flattened plain-text dump (~5 MB cap) β€” paste into any LLM's context | | `/graph.jsonld` | Schema.org JSON-LD entity/concept/source graph | | `/sitemap.xml` | Standard sitemap with `lastmod` | | `/rss.xml` | RSS 2.0 feed of newest sessions | | `/robots.txt` | AI-friendly robots with llms.txt reference | | `/ai-readme.md` | AI-specific navigation instructions | | `/manifest.json` | Build manifest with SHA-256 hashes + perf budget | Every page also includes an `` HTML comment that AI agents can parse without fetching the separate `.json` sibling. #### Recipe β€” query `graph.jsonld` from your terminal The JSON-LD graph isn't just for crawlers β€” you can ask quick questions about your wiki without leaving the shell. Example: print every session as a tree, grouped by project: ```bash python3 examples/scripts/tree_from_graph.py ``` Output: ``` πŸ“š 8 sessions across 3 projects (site/graph.jsonld v1.3.0) llmwiki/ β”œβ”€β”€ demo-blog-engine/ (4 sessions) β”‚ β”œβ”€β”€ 2026-03-12 scaffolding-the-rust-blog-engine β”‚ β”œβ”€β”€ 2026-03-18 adding-syntax-highlighting β”‚ β”œβ”€β”€ 2026-03-25 rss-feed-and-sitemap β”‚ └── 2026-04-01 dark-mode-toggle β”œβ”€β”€ demo-ml-pipeline/ (2 sessions) β”‚ β”œβ”€β”€ 2026-01-20 training-data-pipeline β”‚ └── 2026-02-02 model-training-loop └── demo-todo-api/ (2 sessions) β”œβ”€β”€ 2026-02-08 fastapi-project-bootstrap └── 2026-02-15 adding-oauth-login ``` The full script is **stdlib-only** at [`examples/scripts/tree_from_graph.py`](examples/scripts/tree_from_graph.py). Same recipe pattern works for any aggregation question β€” count sessions per model, find the largest project by token usage, list every entity that appears in 3+ sessions, etc. The graph is yours to slice. ### Quality & governance (v1.0) - **4-factor confidence scoring** β€” source count, source quality, recency, cross-references; with Ebbinghaus-inspired decay per content-type - **5-state lifecycle machine** β€” draft β†’ reviewed β†’ verified β†’ stale β†’ archived with 90-day auto-stale - **16 lint rules** β€” 8 structural (frontmatter, link integrity, orphans, freshness, duplicates, index sync…) + 3 LLM-powered (contradictions, claim verification, summary accuracy) + stale_candidates (#51) + tags_topics_convention (#302) + stale_reference_detection (#303) + frontmatter_count_consistency (#378) + tools_consistency (#378) - **Auto Dream** β€” MEMORY.md consolidation after 24h + 5 sessions: resolve relative dates, prune outdated, 200-line cap - **9 navigation files** β€” CLAUDE.md, AGENTS.md, MEMORY.md, SOUL.md, CRITICAL_FACTS.md, hints.md, hot.md + per-project hot caches ### Obsidian-native experience (v1.0) - **`link-obsidian` CLI** β€” symlinks the whole project into an Obsidian vault; graph view + backlinks + full-text search just work - **Dataview dashboard** β€” 10 ready-to-use queries (recently updated, by confidence, by lifecycle, by project, by entity type, open questions, stale pages) - **Templater templates** β€” 4 templates for source/entity/concept/synthesis pages, seeded with confidence + lifecycle + today's date - **Category pages** β€” tag-based index pages in both Dataview (Obsidian) and static markdown (HTML) modes - **Integration guide** β€” [`docs/obsidian-integration.md`](docs/obsidian-integration.md) covers 6 recommended plugins with per-plugin configs ### Automation - **SessionStart hook** β€” auto-syncs new sessions in the background on every Claude Code launch - **Auto-build on sync** β€” `/wiki-sync` triggers `/wiki-build` (configurable; default on) - **One-shot pipeline** β€” `llmwiki all` runs build β†’ graph β†’ export β†’ lint in a single command (`--strict` for CI) - **MCP server** β€” 12 production tools (query, search, list, read, lint, sync, export, + confidence, lifecycle, dashboard, entity search, category browse) queryable from any MCP client (Claude Desktop, Cline, Cursor, ChatGPT desktop) - **Pending ingest queue** β€” SessionStart hook converts + queues; `/wiki-sync` processes queue - **No servers, no database, no npm** β€” Python stdlib + `markdown`. Syntax highlighting loads from a highlight.js CDN at view time. ## Tutorial β€” every command in 90 seconds A guided tour. Run these in order and you'll have a fully working wiki at `http://127.0.0.1:8765/` by the end. Each command is idempotent and prints what it did. A scripted recording of the same flow ships at [`docs/videos/cli-tutorial.gif`](docs/videos/cli-tutorial.gif) (31 seconds against an 8-session sandbox). The reproducible source is [`docs/videos/cli-tutorial.tape`](docs/videos/cli-tutorial.tape) β€” re-render anytime with `vhs docs/videos/cli-tutorial.tape`. ```bash # 1. One-time scaffold (β‰ˆ1 sec). Creates raw/, wiki/, site/, seed nav files. llmwiki init # 2. Pull in your sessions (β‰ˆ1 sec / 100 sessions). Walks every adapter # that's "available" on this machine (Claude Code, Codex CLI, Cursor, # Gemini, Obsidian, Copilot Chat / CLI), converts new .jsonl files to # raw/sessions/*.md, then runs build + lint by default. llmwiki sync # 3. Compile the static HTML site (β‰ˆ3 sec on a 500-session corpus). # Already runs as part of `sync`; call directly when you're iterating # on a wiki/ page and don't need a fresh sync. llmwiki build # 4. Browse it locally. Cmd+K opens the search palette; / focuses the # filter bar on /sessions/. Press Ctrl+C to stop. llmwiki serve # 5. (Optional) Generate the knowledge graph + AI-consumable exports. # `all` runs build β†’ graph β†’ export β†’ lint in one shot. llmwiki graph llmwiki export all llmwiki all # one-shot equivalent of build + graph + export + lint ``` That's the entire happy path. Two more commands you'll reach for occasionally: ```bash # Inspect what's installed + configured. Prints a per-adapter table: # (available: yes/no, configured: yes/no, session-store path). llmwiki adapters # Lint the wiki. 16 rules β€” broken wikilinks, orphaned pages, stale # summaries, duplicate detection, freshness, missing entities, etc. # Runs as part of `sync` by default; call directly for a one-shot check. llmwiki lint ``` Three optional flags you'll discover later: - `--adapter ` β€” limit `sync` to one adapter (e.g. `--adapter claude_code`) - `--vault PATH` β€” write into an Obsidian / Logseq vault overlay instead of `wiki/` (#54) - `--synthesize` β€” call out to a local Claude / Ollama backend during `build` for an LLM-generated overview page Each subcommand has its own `--help` with the rest. The CLI reference table below is the full list. ## How it works ``` β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚ ~/.claude/projects/*/*.jsonl β”‚ ← Claude Code sessions β”‚ ~/.codex/sessions/**/*.jsonl β”‚ ← Codex CLI sessions β”‚ ~/Library/.../Cursor/workspaceS… β”‚ ← Cursor β”‚ ~/Documents/Obsidian Vault/ β”‚ ← Obsidian β”‚ ~/.gemini/ β”‚ ← Gemini CLI β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β”‚ β–Ό python3 -m llmwiki sync β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚ raw/sessions// β”‚ ← immutable markdown (Karpathy layer 1) β”‚ 2026-04-08-.md β”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β”‚ β–Ό /wiki-ingest (your coding agent) β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚ wiki/sources/.md β”‚ ← LLM-generated wiki (Karpathy layer 2) β”‚ wiki/entities/.md β”‚ β”‚ wiki/concepts/.md β”‚ β”‚ wiki/syntheses/.md β”‚ β”‚ wiki/comparisons/.md β”‚ β”‚ wiki/questions/.md β”‚ β”‚ wiki/index.md, overview.md, log.md β”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β”‚ β–Ό python3 -m llmwiki build β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚ site/ β”‚ ← static HTML + AI exports β”‚ β”œβ”€β”€ index.html, style.css, ... β”‚ β”‚ β”œβ”€β”€ sessions//.html β”‚ β”‚ β”œβ”€β”€ sessions//.txt β”‚ (AI sibling) β”‚ β”œβ”€β”€ sessions//.json β”‚ (AI sibling) β”‚ β”œβ”€β”€ llms.txt, llms-full.txt β”‚ β”‚ β”œβ”€β”€ graph.jsonld β”‚ β”‚ β”œβ”€β”€ sitemap.xml, rss.xml β”‚ β”‚ β”œβ”€β”€ robots.txt, ai-readme.md β”‚ β”‚ β”œβ”€β”€ manifest.json β”‚ β”‚ └── search-index.json β”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ ``` See [docs/architecture.md](docs/architecture.md) for the full 3-layer Karpathy + 8-layer build breakdown. ## Documentation Full production documentation lives under [`docs/`](docs/). The editorial hub is **[`docs/index.md`](docs/index.md)** β€” tutorials, per-agent guides, reference, and deployment, all in one place. **Start here:** | Goal | Read | |---|---| | Install and build your first site in 10 minutes | [Tutorial 01 β†’ 02](docs/tutorials/01-installation.md) | | Use llmwiki with Claude Code | [Tutorial 03](docs/tutorials/03-use-with-claude-code.md) | | Use llmwiki with Codex CLI | [Tutorial 04](docs/tutorials/04-use-with-codex-cli.md) | | Query / lint / review your wiki daily | [Tutorial 05](docs/tutorials/05-querying-your-wiki.md) | | Point llmwiki at an existing Obsidian / Logseq vault | [Tutorial 06](docs/tutorials/06-bring-your-obsidian-vault.md) | | See four real end-to-end workflows | [Tutorial 07](docs/tutorials/07-example-workflows.md) | Contributing to docs? See the **[style guide](docs/style-guide.md)**. ## Install ### macOS / Linux ```bash git clone https://github.com/Pratiyush/llm-wiki.git cd llm-wiki ./setup.sh ``` ### Windows ```cmd git clone https://github.com/Pratiyush/llm-wiki.git cd llm-wiki setup.bat ``` ### With pip (v0.3+) ```bash pip install -e . # basic β€” everything you need pip install -e '.[graph]' # + graphifyy AI-powered graph engine pip install -e '.[dev]' # + pytest + ruff pip install -e '.[e2e]' # + Playwright + pytest-bdd + pytest-html (E2E) pip install -e '.[all]' # graph + everything ``` Syntax highlighting is now powered by [highlight.js](https://highlightjs.org/), loaded from a CDN at view time β€” no optional deps required. ### What setup does 1. Creates `raw/`, `wiki/`, `site/` data directories 2. Installs the `llmwiki` Python package in-place 3. Detects your coding agents and enables matching adapters 4. Optionally offers to install the `SessionStart` hook into `~/.claude/settings.json` for auto-sync 5. Runs a first sync so you see output immediately ## For maintainers Running the project? The governance scaffold lives under [`docs/maintainers/`](docs/maintainers) and is loaded by a dedicated skill: | File | What it's for | |---|---| | [`CONTRIBUTING.md`](CONTRIBUTING.md) | Short rules for contributors β€” read this first | | [`CODE_OF_CONDUCT.md`](CODE_OF_CONDUCT.md) | Contributor Covenant 2.1 | | [`SECURITY.md`](SECURITY.md) | Disclosure process for redaction bugs, XSS, data leaks | | [`docs/maintainers/ARCHITECTURE.md`](docs/maintainers/ARCHITECTURE.md) | One-page system diagram + layer boundaries + what NOT to add | | [`docs/maintainers/REVIEW_CHECKLIST.md`](docs/maintainers/REVIEW_CHECKLIST.md) | Canonical code-review criteria | | [`docs/maintainers/RELEASE_PROCESS.md`](docs/maintainers/RELEASE_P … ## Source & license This open-source MCP server is cataloged on AgentStack and links to its original source β€” we do not rehost the code. - **Author:** [Pratiyush](https://github.com/Pratiyush) - **Source:** [Pratiyush/llm-wiki](https://github.com/Pratiyush/llm-wiki) - **License:** MIT Install and usage instructions live in the source repository linked above. ## Pricing - **Free** β€” Free ## Security capabilities Automated source analysis of v0.1.0 β€” what this tool can access: - **Network access:** no - **Filesystem access:** no - **Shell / process execution:** no - **Environment & secrets:** no - **Dynamic code execution:** no *"Yes" means the capability is present in the source β€” more access means more to trust, not that it is unsafe.* ## Versions - **0.1.0** β€” security scan: passed β€” Imported from the upstream source. ## Links - Listing page: https://agentstack.voostack.com/l/mcp-pratiyush-llm-wiki - Seller: https://agentstack.voostack.com/s/pratiyush - Browse the marketplace: https://agentstack.voostack.com/browse --- Listed on AgentStack β€” the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.