Feature: Better session organization (folders/projects/tags) like Claude.ai

Problem

Claude Code stores conversations as JSONL files in ~/.claude/projects/<project-hash>/, but there's no way to organize them. As my sessions grow into the hundreds (I have 290+ right now), finding a past conversation becomes increasingly painful:

• No folders/projects to group related sessions together
• No custom tags or labels to categorize sessions (e.g., by sprint, feature, bug)
• No full-text search across conversation content in the /resume picker
• No session naming by default — I'm stuck scrolling through first prompts trying to recognize what I worked on
• Only filter available is by git branch, which doesn't map 1:1 to conceptual "projects"

Compared to Claude.ai which has Projects/Folders for organizing conversations, Claude Code feels flat and un-browsable once you have more than a few dozen sessions.

Use Cases

• "Show me all sessions from last week's security sprint" → can't do this cleanly today
• "Find the conversation where we discussed rate limiting" → have to remember the branch or scroll forever
• "Group all sessions related to this epic" → impossible without renaming every one manually
• "Tag this session as bug-fix so I can find it later" → not supported

Proposed Features (in priority order)

1. Named projects/folders that can contain sessions (like Claude.ai projects)
2. Custom tags/labels per session
3. Full-text search across session content from the /resume picker
4. Persistent sidecar metadata (separate from JSONL so Claude's own files stay clean)
5. Better default titles — strip command metadata (e.g. <command-message>… currently shows as session title)
6. Three-pane browser like macOS Finder: projects / sessions / preview

Workaround (meanwhile)

I built an npm CLI tool as a stopgap: claude-code-sessions

Install:

npm install -g claude-code-sessions

Then run:

claude-sessions          # opens an interactive three-pane TUI browser
claude-sessions search "rate limiting"    # full-text search
claude-sessions project create "Sprint 5" # create a project

It reads Claude's ~/.claude/projects/ (strictly read-only) and stores tags/projects in a sidecar file at ~/.claude-sessions/metadata.json. Source and docs: https://github.com/nikkhildev/claude-sessions

It's working well for me, but I'd love to see this as a first-class Claude Code feature so everyone benefits — especially the full-text search in /resume and the projects/folders organization.

Questions for the team

1. Is session organization on the roadmap? Any plans for projects/tags?
2. Would you accept a community contribution, or is this something you'd prefer to design in-house?
3. Is the sessions-index.json schema stable enough to build tooling against, or is it expected to change?

Thanks!