By Snowtumb
LLM Wiki for codebases — uses Serena as the storage and CRUD layer to build a persistent, interlinked knowledge base of your codebase's architecture, patterns, decisions, and component relationships. All wiki pages are Serena memories. Inspired by Karpathy's LLM Wiki pattern.
This skill should be used when the user asks to "enrich the wiki", "add external context", "find relevant docs", "research X for the wiki", "add documentation references", "what do the official docs say about X", "find articles about our approach to X", "add best practices to the wiki", "wiki enrich", "external context", or "web research for wiki". Also use when the user wants to understand how their technology choices compare to best practices. Searches the web and writes external context pages as Serena memories.
This skill should be used when the user asks to "check the wiki", "lint the wiki", "wiki health check", "find stale wiki pages", "audit the wiki", "are there outdated pages", "wiki quality", "fix broken links in the wiki", "wiki maintenance", "wiki lint", "is the wiki accurate", "verify the wiki", or mentions wanting to ensure wiki quality. Performs a comprehensive health check on the wiki stored in Serena memories — finds stale pages, orphans, missing cross-references, and content quality issues.
This skill should be used when the user asks questions about their codebase such as "why did we build X this way", "how does X work", "what is the architecture of X", "explain the pattern for X", "what depends on X", "how is X connected to Y", "what decisions led to X", "what's the reasoning behind X", "show me how X is structured", or any question that the wiki might answer. Also use when the user says "check the wiki", "wiki query", "search the wiki", "what does the wiki say about X", "look it up in the wiki", or "find X in the wiki". Reads wiki pages from Serena memories and synthesizes answers.
This skill should be used when the user asks to "scan the codebase", "build the wiki", "populate the wiki", "deep scan", "document the codebase", "generate wiki pages", "analyze the codebase for the wiki", "fill in the wiki", "wiki scan", or "wiki build". Also use after wiki-setup when the user wants to populate the wiki with codebase knowledge. Performs a deep semantic analysis of the entire codebase using Serena and writes all wiki pages as Serena memories. Dispatches parallel agents for maximum speed.
This skill should be used when the user asks to "set up the wiki", "initialize the wiki", "create a wiki for this project", "start documenting this codebase", "set up codebase documentation", "init wiki", "wiki setup", or mentions wanting a persistent knowledge base for their codebase. Creates the wiki memory structure inside Serena with schema, index, log, and a lightweight overview. This is the entry point — run this before any other wiki skill.
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 that builds and maintains a persistent, interlinked wiki of your codebase — powered by Serena.
Instead of rediscovering how your code works every session, Serena Wiki incrementally builds a structured knowledge base — architecture decisions, module relationships, code patterns, and the reasoning behind them. The wiki compounds over time: every scan, question, and update makes it richer.
All wiki pages are stored as Serena memories — Serena is the storage layer, the code analysis engine, and the CRUD interface. Everything goes through Serena.
Inspired by Karpathy's LLM Wiki pattern.
Serena is required. This plugin will not work without it. Install Serena first:
/plugin install serena@claude-plugins-official
# Add the marketplace
/plugin marketplace add Snowtumb/Karpathy-serena-wiki
# Install the plugin
/plugin install serena-wiki@Karpathy-serena-wiki
# Reload
/reload-plugins
# 1. Initialize the wiki (creates Serena memories)
/wiki setup
# 2. Deep-scan your codebase
/wiki scan
# 3. Ask questions — wiki memories are consulted first
"How does authentication work in this project?"
"Why did we choose PostgreSQL?"
"What depends on the auth module?"
# 4. After making changes, update the wiki
/wiki update
# 5. Periodically check wiki health
/wiki lint
# 6. Add external documentation context
/wiki enrich
Every wiki page is a Serena memory. Serena handles everything:
| Operation | Serena Tool |
|---|---|
| Create a wiki page | write_memory |
| Read a wiki page | read_memory |
| Update a wiki page | edit_memory |
| Delete a wiki page | delete_memory |
| Browse the wiki | list_memories (topic: "wiki") |
| Search wiki + code | search_for_pattern |
| Analyze code symbols | get_symbols_overview, find_symbol |
| Trace dependencies | find_referencing_symbols |
Wiki pages are organized with / topic separators:
wiki/
├── wiki/schema # Conventions (self-documenting)
├── wiki/index # Categorized catalog of all pages
├── wiki/log # Chronological record of operations
├── wiki/overview # Project purpose, tech stack, layout
├── wiki/architecture/system-design
├── wiki/architecture/data-flow
├── wiki/modules/auth
├── wiki/modules/api
├── wiki/patterns/error-handling
├── wiki/patterns/code-conventions
├── wiki/decisions/database-choice
└── wiki/external/react-docs
/wiki setup — InitializeCreates wiki memories: schema, index, log, and a lightweight overview.
Triggers: "set up the wiki", "initialize wiki", "create a wiki for this project"
/wiki scan — Deep Codebase Analysis6-phase scan using Serena's semantic analysis:
list_dirfind_referencing_symbolsget_symbols_overviewsearch_for_patternTriggers: "scan the codebase", "build the wiki", "document the codebase"
wiki-query — Search and AskReads wiki memories first, investigates code with Serena only if the wiki has gaps. Cites wiki pages. Suggests capturing new knowledge.
Triggers: "how does X work?", "why did we build X?", "what depends on X?"
/wiki update — Sync After ChangesThree modes: explicit, diff-based, conversational. Updates wiki memories via edit_memory.
Triggers: "update the wiki", "document what we just did", "record this decision"
/wiki lint — Health CheckChecks: staleness, orphan memories, missing cross-references, empty sections, coverage gaps. Offers auto-fix for mechanical issues.
Triggers: "lint the wiki", "wiki health check", "is the wiki up to date?"
/wiki enrich — External ContextSearches the web, writes external context pages as Serena memories, cross-references with internal pages.
Triggers: "enrich the wiki", "find relevant docs", "what do the official docs say?"
npx claudepluginhub snowtumb/karpathy-serena-wiki --plugin serena-wikiBootstrap a wiki-first knowledge system into any project.
Local-first memory server — hybrid BM25+vector search, vault management, lint, and launchd lifecycle for project knowledge.
AI-powered knowledge base management - Capture conversation learnings, maintain topic-specific KB files, Obsidian-compatible knowledge graph, structured dynamic context loading, and institutional knowledge in CLAUDE.md
Query GitHub repository wikis via DeepWiki CLI without MCP token overhead.
Optimized file search, semantic indexing, and persistent memory for Claude Code — with optional sync to a self-hosted web dashboard
Unified context retrieval from Forgetful Memory, Context7 docs, and Serena symbol analysis