Zikaron (זיכרון) - Knowledge Pipeline

Zikaron indexes Claude Code conversation history and markdown files into a searchable vector database. Query past solutions, code patterns, and debugging sessions.

Prerequisites

bash

# Ollama must be running for embeddings
ollama serve &
ollama pull nomic-embed-text

Quick Reference

Index Conversations

bash

# Index all Claude Code conversations
zikaron index

# Index specific project only
zikaron index -p project-name

Index Markdown Files

bash

# Index learnings, skills, CLAUDE.md files, research docs
zikaron index-md ~/Gits/claude-golem/docs.local/learnings
zikaron index-md ~/Gits/claude-golem/skills
zikaron index-md ~/path/to/any/markdown/directory

Content types by path:

Path Pattern	Type	Value
`/learnings/`	learning	HIGH
`/skills/`	skill	HIGH
`CLAUDE.md`	project_config	HIGH
`/research/`	research	HIGH
`/prd/*`	prd_archive	MEDIUM
`/verification`	verification	LOW
Default	documentation	MEDIUM

Search

bash

# Basic search
zikaron search "how did I implement authentication"

# Filter by project
zikaron search "error handling" -p my-project

# Filter by content type
zikaron search "debugging" -t learning

# More results
zikaron search "api design" -n 20

Stats & Management

bash

# View knowledge base stats
zikaron stats

# Clear entire database (careful!)
zikaron clear -y

Storage Location

Database: ~/.local/share/zikaron/chromadb

Check size: du -sh ~/.local/share/zikaron/chromadb

MCP Integration

Add to ~/.claude/settings.json:

json

{
  "mcpServers": {
    "zikaron": {
      "command": "zikaron-mcp"
    }
  }
}

Then Claude Code can query directly: "Search my past conversations for authentication patterns"

Troubleshooting

Schema Errors

If you see "no such column" errors after upgrading:

bash

rm -rf ~/.local/share/zikaron/chromadb
zikaron index  # Re-index from scratch

Ollama Not Running

bash

# Check if running
curl http://localhost:11434/api/tags

# Start if needed
ollama serve &

Large Database Size

ChromaDB 0.5.x had bloat issues. Zikaron pins to 0.4.x for stability. If database is huge:

bash

rm -rf ~/.local/share/zikaron/chromadb
zikaron index  # Should be much smaller

Architecture

code

~/.claude/projects/     →  Pipeline  →  ChromaDB  →  CLI / MCP
   (JSONL files)           (5 stages)    (vectors)    (query)

Stages:
1. Extract - Parse JSONL, dedupe system prompts
2. Classify - Identify content type (code, errors, messages)
3. Chunk - AST-aware splitting for code, header-based for markdown
4. Embed - Generate vectors via Ollama (nomic-embed-text)
5. Index - Store in ChromaDB with metadata

When to Use

•Finding past solutions: "How did I handle that API rate limiting before?"
•Debugging patterns: "What error messages have I seen with this library?"
•Code reuse: "Find my previous implementation of pagination"
•Learning recall: "What did I learn about jq escaping?"