LLM Wiki — a second brain for AI agents

Turn PDFs, articles, transcripts, and notes into a shared wiki that your AI agents can search, cite, and keep up to date. Add a source once. Ask questions later. Keep the useful answers.

Works with Claude Code, Codex, Cursor, Gemini CLI, OpenCode, OpenClaw, Pi, and OMP. Read the documentation.

What is LLM Wiki?

AI agents are good at the task in front of them, but a new session starts with limited context. LLM Wiki gives them a shared memory that lives inside your project.

When you add a source, the agent turns it into linked Markdown pages. Later, it can find the right section and answer with citations. Useful answers can be saved back into the wiki, so the knowledge grows instead of being rebuilt from scratch.

Everything canonical stays in readable Markdown. Default semantic search is local—no hosted vector database or embedding service.

What's new in v3.0.0

• Local semantic search by default. FastEmbed runs BAAI/bge-small-en-v1.5 on-device; sqlite-vec stores derived vectors in each wiki's cache.
• Hybrid precision and recall. BM25 exact-term ranks and semantic ranks are fused with RRF.
• Incremental indexing. Content hashes limit re-embedding to new or changed sections and remove deleted sections.
• Zero provider surface. No OpenAI-compatible endpoint, API key, remote text transfer, provider consent, or per-query charge.
• Safe lexical escape hatch. --no-embed stays pure-Python BM25; local backend failures fall back to it automatically.
• No content migration. Existing Markdown remains valid and old embeddings.jsonl caches are simply ignored.

Why use it?

• Stop repeating project context. Your agent can read the knowledge you already collected.
• Use the same wiki across agents. The files are plain Markdown, not tied to one model or tool.
• Trace every answer. Citations point back to the wiki page and original source.
• Keep the wiki healthy. The agent updates links, summaries, and contradictions as new sources arrive.
• Stay in control. Your pages remain readable, editable, and versionable.

LLM Wiki works best for knowledge that grows over time: research, meeting notes, customer calls, papers, articles, and project decisions. Use a regular database when your main problem is structured records and transactions.

Installation

uv is the only prerequisite: it creates the pinned script environments on every supported platform. Install it using the official uv instructions before running /wiki:init or the natural-language equivalent.

Plugin/skill installation copies all agent-facing commands and bundled tools. Wiki initialization then performs mandatory runtime setup: FastEmbed 0.8.0, sqlite-vec 0.1.9, PyYAML 6.0.3, the local BAAI/bge-small-en-v1.5 model, the parse cache, and vectors for every existing section. Upgrade runs the same setup again, incrementally synchronizing changed and deleted sections.

Claude Code — full plugin

The native path: the skill, the seven /wiki:* slash commands, and the marketplace manifest all ship in one install.

/plugin marketplace add praneybehl/llm-wiki-plugin
/plugin install llm-wiki@llm-wiki

Once installed, the plugin works in any project — the wiki itself lives in the project's working directory, not in the plugin.

Other coding agents — skill only

The llm-wiki skill uses the standard agentskills.io format, so it installs cleanly into any agent supported by the skills CLI. Pick the --agent flag that matches your setup:

# Install globally so the skill is available across all projects
npx skills add praneybehl/llm-wiki-plugin -a <agent> -g

# Or install into the current project only
npx skills add praneybehl/llm-wiki-plugin -a <agent>

| Agent | --agent value | Invoke via | Scripts run | |-------|-----------------|------------|:-----------:| | Claude Code | claude-code | /wiki:* slash commands (bundled) or natural language | ✅ | | Codex (OpenAI) | codex | /skills or $llm-wiki / natural language | ✅ | | Cursor | cursor | /llm-wiki or natural language | ✅ | | Gemini CLI | gemini-cli | /skills management commands / natural language | ✅ | | OpenCode | opencode | natural language (agent invokes the native skill tool) | ✅ | | OpenClaw | openclaw | auto-exposed as a user command | ✅ | | Pi Agent | pi | /skill:llm-wiki or natural language | ✅ | | OMP ("Oh My Pi") | manual (see below) | natural language (skills auto-surface via skill://) | ✅ |

OpenCode also reads .claude/skills/ and ~/.claude/skills/, so if you already installed the skill for Claude Code you can use it in OpenCode without a second install.

Hermes Agent (Nous Research), OMP ("Oh My Pi"), and other agentskills.io-compatible runtimes that aren't yet in the npx skills registry can still use this skill — clone the repo and symlink or copy skills/llm-wiki/ into the agent's skills directory. Hermes reads from ~/.hermes/skills/; OMP reads managed/user skills from ~/.omp/agent/skills/ and surfaces them via skill://.

git clone https://github.com/praneybehl/llm-wiki-plugin.git
mkdir -p ~/.hermes/skills ~/.omp/agent/skills
ln -s "$(pwd)/llm-wiki-plugin/skills/llm-wiki" ~/.hermes/skills/llm-wiki
ln -s "$(pwd)/llm-wiki-plugin/skills/llm-wiki" ~/.omp/agent/skills/llm-wiki

A few things to know when using the skill outside Claude Code:

• Slash commands are Claude Code-only. The seven /wiki:* commands live in commands/wiki/ as Claude Code plugin manifests. In other agents, invoke the skill by natural language ("add this paper to the wiki", "what does the wiki say about X", "lint the wiki") — the SKILL.md handles the rest.
• All bundled tools are agent-accessible. Every listed agent can invoke setup_wiki.py, hybrid/lexical search, lint, stats, graph lint/extract/query, and initialization through the installed skill. Dependency-bearing scripts carry pinned PEP 723 metadata and run with uv run --script; initialization and upgrade verify the full runtime before reporting readiness.
• The wiki itself is agent-agnostic. It's just a directory of markdown files. You can ingest with one agent and query with another; nothing in wiki/ ties it to a specific runtime.

Quick start

In a project where you want to keep a wiki:

/wiki:init

Already have a wiki from an earlier plugin version? Run /wiki:upgrade instead. Init and upgrade both install and verify the complete pinned local runtime, cache the embedding model, build the parse cache, and synchronize all current vectors. They add missing wiki files idempotently and never overwrite existing pages or SCHEMA.md.

Initialization bootstraps wiki/ and raw/ with SCHEMA.md, index.md, log.md, and a page template, then emits a JSON runtime report. Setup is complete only when it reports "status": "ready".

As part of the same step, the skill will propose wiring the wiki into your project's agent-memory file so the agent remembers it in future sessions without being told. The target file depends on which agent you use: CLAUDE.md for Claude Code, AGENTS.md for Codex / Cursor / OpenCode / Pi / OpenClaw / OMP, GEMINI.md for Gemini CLI. If you run multiple agents in the same project, use AGENTS.md and symlink CLAUDE.md to it. The skill never writes to a memory file without your approval — see skills/llm-wiki/references/agent-memory-integration.md for the canonical stanza and a three-line short variant.

Drop your first source into raw/ (a PDF, a markdown clipping from the Obsidian Web Clipper, a transcript, anything textual), then:

/wiki:ingest raw/your-source.pdf

Claude reads the source (chunk-reading if it's large), discusses the takeaways with you, writes a summary page, identifies which existing entity and concept pages it touches, surgically updates them with str_replace, creates new pages for new entities and concepts (each with at least one inbound link), updates the index, and appends to the log.

Once you have a few sources in:

/wiki:query What does my wiki say about X?

Claude reads the index to find candidate pages, reads them, synthesizes an answer with [[wikilink]] citations, and offers to file substantive answers back into wiki/synthesis/ so the exploration compounds.

Periodically, run a health check:

/wiki:lint

This catches orphan pages, broken wikilinks, oversized pages, missing frontmatter, and stale claims. Findings are presented as proposed edits — Claude doesn't rewrite your wiki silently.

Commands

| Command | What it does | |---------|--------------| | /wiki:init | Bootstrap a new wiki structure in the current project (now includes wiki/graph/). | | /wiki:ingest <source> | Process a new source into the wiki; refreshes the graph layer when present. | | /wiki:query <question> | Answer a question from the wiki with citations; consults graph.sqlite for relational questions when available. | | /wiki:lint | Structural and semantic health check; also runs graph lint when wiki/graph/ontology.yaml exists. | | /wiki:stats | Show wiki size, link density, and which scaling threshold the wiki is at. | | /wiki:graph <action> | extract / lint / neighbors / edges / path / facts against the compiled graph layer. | | /wiki:upgrade | Upgrade an existing wiki to the current plugin version (idempotent file ops + walked SCHEMA.md merge). |

You don't have to use the commands — Claude triggers the underlying skill on natural-language requests too ("add this paper to my wiki", "what does the wiki say about diffusion", "lint the wiki"). The commands are there when you want explicit invocation.

Architecture

The wiki has three layers and three operations.

The three layers are raw sources (your immutable source material in raw/), the wiki (LLM-owned markdown pages in wiki/, organized into sources/, entities/, concepts/, synthesis/), and the schema (wiki/SCHEMA.md, the configuration file that documents the conventions for this specific wiki and that you co-evolve with Claude over time).

The three operations are ingest (compile a new source into the wiki, touching ten or fifteen pages in one pass), query (navigate the wiki to answer a question, with the option to file the answer back as a new synthesis page), and lint (periodic health check on structural and semantic integrity).

There is also an optional fourth layer: the graph at wiki/graph/. Pages can declare typed graph: metadata in frontmatter (e.g. founded, proposed, depends_on) with an explicit source-page slug and evidence quote. A bundled extractor compiles every page into nodes.jsonl, edges.jsonl, graph.sqlite, and graph.graphml. Markdown stays canonical — the graph can be deleted and rebuilt at any time. Use /wiki:graph for relational queries that the index alone can't answer cheaply ("what's connected to X", "who proposed Y", "shortest path A → B").

For the full architecture write-up including page types, frontmatter conventions, the rationale behind each design choice, and the workflow procedures, see the reference docs inside the skill at skills/llm-wiki/references/. The graph layer is documented in skills/llm-wiki/references/graph-workflow.md.

How it scales

The single biggest failure mode of the LLM Wiki pattern is the wiki becoming a context bottleneck as it grows. The plugin's design is shaped largely around avoiding that failure.

Atomic pages. Every page is about one concept. Soft cap 400 lines, hard cap 800 lines. The lint script enforces this. Any single page read is bounded, so even at a thousand pages a query that reads five pages stays well within context.

Sharded indexes. When `wiki