xstitch

Cross-tool memory for AI agents. Stitch context across Cursor, Claude Code, Codex, Gemini CLI, Windsurf, Copilot, and 9+ AI tools. Preserves decisions, failed experiments & warnings across sessions. Zero dependencies, MCP + instruction files, BM25 relevance search. Never re-explain your work.

The Problem

You're deep in a task in Cursor. You hit the token limit. You open Claude Code to continue.

The new agent has zero memory.

It doesn't know you tried approach X and it failed. It doesn't know you chose library Y for a reason. It wastes 15 minutes rediscovering what took you 45 minutes to figure out.

This happens constantly:

• Tool switching kills context — switch from Cursor to Claude Code to Codex, and each agent starts blind
• Context summarization loses nuance — auto-summaries drop the failed experiments, rejected alternatives, and "don't try this" warnings
• Session crash = square one — terminal closes, laptop restarts, and the agent has to re-read every file, re-check every diff
• Repeating the same mistakes — Agent B tries the same approach Agent A already spent 45 minutes discovering doesn't work
• Token-expensive rediscovery — every new session burns tokens re-reading files, re-checking diffs, and re-building understanding that already existed
• No git, no memory — existing tools only capture context on commits, but most AI work happens between commits while exploring and debugging