prashantpandey-creator

Meditate Sessions — Communication skill for Claude Code

Communication community

Read your Claude Code sessions, find the threads inside, resume one.

How to install Meditate Sessions

This entry records only its repository, not the path inside it, so there is no exact command to give. Open prashantpandey-creator/meditate-sessions and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Meditate Sessions does

Read your Claude Code sessions, find the threads inside, resume one. Streams transcripts without loading them; writes a paste-able continuation chat per live thread. No dependencies outside the standard library.

Alternatives in Communication

Claude Skill Social Post — A Claude Code skill by Hao (駱君昊) that learns your Facebook voice and auto-posts to FB / IG / Threads / X with 625 ★
YouTube To Ebook — Claude skill for turning YouTube transcripts from your favorite channels into EPUB ebooks, delivered to your e 426 ★
Mcptube — Transform YouTube videos into a compounding knowledge base with transcripts, vision analysis, and agentic sear 152 ★

README

meditate

**One long Claude Code session becomes a 100 MB tangle of unrelated work. `meditate` reads it without loading it, finds the separate threads inside, and hands you a paste-able prompt to resume any one of them.**

curl -fsSL https://raw.githubusercontent.com/prashantpandey-creator/meditate-sessions/main/get.sh | sh

Then:

meditate sessions

329 sessions across 17 projects

  sprawl   126  117.1MB    83u  ch:0   [vedic-puran]  Game work elements resume
  sprawl    90   39.3MB   296u  ch:12  [vedic-puran]  chart engine + pricing
  sprawl    78   29.2MB   238u  ch:12  [vedic-puran]  reader latency

Sprawl is how many distinct threads are tangled in one session. The top row is the one you keep scrolling through to find where you were.

The commands


`meditate sessions`	every session, ranked by how tangled it is
`meditate split`	one session, broken into its threads
`meditate threads`	what is still open across everything you split
`meditate open`	a Terminal per live thread, cd'd and prompted (macOS)
`meditate archive`	what's finished and can be set down (dry run)
`meditate repo`	the optional repo lens
`meditate test`	run the suite

How it reads a 100 MB transcript

By streaming it. A transcript never enters a context window — there is no model call anywhere in this tool, and no network call at all. Each session comes back as a capped record: title, sprawl, where the topic changed, the human intents with tool noise stripped, and the files it touched.

That is also why it is fast and why it costs nothing to run.

What it will not do

It never moves, renames or deletes a transcript or a project file. Every delete site in the source is enumerated in `test_product.py` and a new one fails the suite. It writes to two places: your project's memory directory and `~/.meditation/`. `archive` is a dry run unless you pass `--apply`, and archiving is reversible.

`install.sh`