Community Tool: ccs — Fast CLI for searching Claude Code session transcripts

I built a small CLI tool to search across Claude Code session history — something I've wanted since #6912 was filed.

What it does

ccs (Claude Code Session Search) searches your .jsonl session files using ripgrep + Python:

ccs "RabbitMQ"                        # search all sessions
ccs "backup script" --user            # only in your messages
ccs "docker compose" --assistant      # only in Claude's responses
ccs "webhook" --project my-app        # filter by project
ccs "deploy" --since 2025-03-01       # date filter

Key features

• Fast — ripgrep for file-level matching (~70ms across 1000+ sessions / 2GB), then Python for structured filtering
• Smart --user filter — distinguishes actual human-typed messages from tool results (which are also stored as type: "user" records in the JSONL format)
• Zero dependencies — Python stdlib only. Auto-detects ripgrep (system install or Claude Code's bundled binary via ARGV0 trick), falls back to grep
• --verbose shows claude --resume <id> commands for each matching session
• --json for programmatic output

Output example

5 matches in 3 sessions

2025-03-28  migrate-webpay-checkout  [~/Sites/magento]
  [claude] 01:46  ...RabbitMQ exchange recreation after fresh container start...

2025-03-20  switch-production-vulcan-to-leti  [~/Sites/magento]
  [you] 14:00  ...Redis, OpenSearch, and RabbitMQ connection settings should match...

Install

git clone https://github.com/ignaciomella/claude-session-search.git
ln -s "$(pwd)/claude-session-search/ccs" ~/.local/bin/ccs

Repo also includes a Claude Code skill definition for natural language queries ("search my sessions for where I mentioned X").

Sharing in case others find it useful. Feedback welcome!