claudestrophobic

skill
Security Audit
Warn
Health Warn
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 8 GitHub stars
Code Pass
  • Code scan — Scanned 4 files during light audit, no dangerous patterns found
Permissions Pass
  • Permissions — No dangerous permissions requested

No AI report is available for this listing yet.

SUMMARY

Brings conversation management back.

README.md

claudestrophobic

Because 064ddd26-9a3f-... is not a name.

Stars

Claude Code Plugin MIT License Zero context overhead
demo

A Claude Code plugin that brings conversation management back — list, delete, and prune chats, and clean up whole projects, without leaving the terminal. Built as skills, not MCP tools, so they occupy zero tokens in your context window until you invoke them.

Install

/plugin marketplace add urav06/claudestrophobic
/plugin install claudestrophobic@claudestrophobic-marketplace

Usage

/sessions — chats in the project you're in:

/sessions                              # list this project's chats
/sessions delete the one about auth    # delete by description (resolved to a chat)
/sessions delete 064ddd26              # …or by UUID prefix
/sessions prune --older 2w             # delete chats older than 2 weeks
/sessions browse                       # open this project's folder

/projects — the whole map, managed from outside:

/projects                              # list every project: chats, size, live / orphaned
/projects nuke them-ctf                # preview removing a project you're done with
/projects nuke them-ctf --confirm      # …then nuke it: chats, memory, history, dir → Trash

Why

On claude.ai, your conversations live in a sidebar. Named, browsable, deletable. In Claude Code, they become invisible UUID files under ~/.claude/projects/. There is no built-in way to delete them. This has been one of the most requested features for over 10 months — with no response from Anthropic.

This brings that visibility back.

How it works

  • Names resolved from chat transcript metadata — /rename titles, auto-generated titles, or first message. Same priority chain Claude Code uses internally.
  • Deletion informed by Claude Code's surfaced internals — cleans transcripts, subagent data, file history, session environment, and history rows. Nothing left dangling.
  • Projects discovered by their real working directory, read losslessly from the transcript (not the lossy encoded folder name) — so chats orphaned by a deleted folder are detected and reclaimable. Claude Code never cleans these up on its own.
  • nuke is total and deliberate: it takes a finished project's chats, memory, history, and directory in one go. It refuses the project you're currently in and any project with an active chat.
  • Safety — active chats detected from lock files and protected. History rewrites are atomic (a concurrent session never sees a half-written file). Everything goes to the system Trash where available, rm as fallback. Cross-platform.
  • Overhead — only loads into context when you invoke a skill. Zero tokens otherwise.

Reviews (0)

No results found