We've added a SessionStart hook to Threadlinking. Now when you open Claude Code, it runs "threadlinking context" and shows threads/pending files filtered to your current project. Claude gets project context before you even ask.

Here's what the Threadlinking flow looks like when you ask Claude to run "threadlinking list":

Why "threadlinking"? A thread is a continuous line of reasoning across sessions. Linking connects your files to that reasoning. Threadlinking = linking your code, files, and documents to the thread of thought where they originated.

Threadlinking is like git for decisions. It preserves your Claude Code conversation snippets and decision-making rationale locally, so Claude can easily retrieve context about the decisions you made when you're working together, across sessions.

Threadlinking is becoming a plugin with MCP server and skills for Claude Code. It preserves context about your cognitive work and decisionmaking, with snippets from conversations at decision points. So far testing it, the features work so much better than just the CLI tool.

This sums up what the tool does for context preservation in Claude Code. Looking for more beta testers DM feedback on whether it works for you (or open issues or PRs in gh).

New feature. Telling Claude via the global dot md "remember to save context" didn't work well, because it forgets. threadlinking v1.1.0 installs a hook that tracks files automatically. At the session start, Claude sees what's untracked and then prompts you to save context.

I first created Threadlinking because I was generating a ton of projects via Claude/ChatGPT, associated repos and files. I couldn't keep track. I needed context preservation for cognitive work. Now that I do everything in Claude Code, I've revised it for that ecosystem: