Cloudctx

CloudCtx

Persistent memory for [Claude Code](https://docs.anthropic.com/en/docs/claude-code). One command, full recall.

The problem

Claude Code forgets everything between sessions. When context gets long, it compacts — and forgets even more. You end up re-explaining your project, your preferences, and your past decisions. Every. Single. Time.

The fix

CloudCtx gives Claude Code permanent memory. Every conversation is indexed into a local SQLite database. Your agent can search across months of history instantly, survives compaction without losing context, and you can jump back into any saved thread with a single command.

# Prereq: Bun (https://bun.sh)
curl -fsSL https://bun.sh/install | bash

npm install -g cloudctx
cloudctx init

That's it. Zero config. Works immediately.

---

Three features that matter

1. Persistent memory

Every message from every Claude Code session is indexed into a local SQLite database with FTS5 full-text search. Your agent can recall past errors, decisions, and solutions across all projects and sessions.

Claude searches it automatically (CloudCtx adds instructions to your `CLAUDE.md`), or you can search manually:

# Full-text search across all conversations
cloudctx query "redis connection timeout"

# Raw SQL for complex queries
cloudctx sql "SELECT type, substr(content,1,200), timestamp FROM messages WHERE content LIKE '%deploy%' ORDER BY timestamp DESC LIMIT 5"

The database grows with you. Months of conversations, tens of thousands of messages — SQLite handles it without breaking a sweat.

2. Compaction recovery

When Claude Code compresses your context (manually via `/compact` or automatically when the conversation gets long), important details disappear. CloudCtx detects compaction and automatically re-injects the last 40 messages from your database back into the conversation.

The result: your agent picks up where it left off instead of starting from scratch. No more "I don't have conte