resume-resume

New free tool we're dropping.

Paste a dead chat back to life — in milliseconds

Your session crashed; the terminal still shows the tail of it. Copy any chunk of that text, run cr, paste. It finds the exact chat that text came from and reopens it — no session id, no scrolling a list. Indexed full-text search across your whole history, and unlike grep/rg it ranks and verifies the match — so it opens the right chat (0% wrong in testing), not the first hit.

Related MCP server: Session Buddy

MCP Server

Add it to Claude Code and every session on your machine can search, read, and merge your full session history — in plain English. Indexes both Claude Code and Codex CLI sessions, so one query reaches across both tools.

pip install resume-resume
claude mcp add resume-resume -- resume-resume-mcp

Or manually in your MCP config:

{
  "mcpServers": {
    "resume-resume": {
      "command": "resume-resume-mcp"
    }
  }
}

We built Eidos, a multi-agent AI system. In our benchmark, Eidos outperformed Claude Opus 4.6 by 3.6x in both accuracy and speed on complex tasks with 15+ reasoning chains. Below, we use Claude Resume to pick up where we left off across multiple sessions.

Finding the benchmark where Eidos beat Claude Opus 4.6

"use resume-resume to find the eidos test where we beat claude"

Searching for a past session in plain English

"use resume-resume to find the latest chats about eidos philosophy docs"

Merging multiple past sessions into this chat

"use claude resume to merge march 14th conversations and Eidos v5 Pipeline Telemetry convo from march 11th into this chat"

Two sessions — one about eidos-philosophy doc changes (Mar 14) and one with a full 28-task strategic plan (Mar 11) — merged into the current conversation with a single command.

MCP Tools

How to Resume

Three ways, fastest first.

1. Paste box — just run cr

Run cr with no arguments and it shows a paste box before the session list. Paste anything and press Enter:

📋 Paste a resume command, session id, or chat text — or press Enter for the list:
›

The third one is the point: your session crashed, the terminal still shows the last thing on screen, you copy a chunk of it, run cr, paste. It finds the chat and reopens it — no id, no hunting.

2. Paste a command directly as an argument

claude --resume <id> fails if you're in the wrong directory. cr fixes that — auto-discovers the project, ensures the right flags, and launches.

cr claude --resume ddf7fc98-6c93-40c8-9444-503d8a716dbf
cr claude --resume ddf7fc98-6c93-40c8-9444-503d8a716dbf --model opus --chrome
cr ddf7fc98-6c93-40c8-9444-503d8a716dbf

It parses the id out of whatever you pasted, searches ~/.claude/projects/*/ for the owning project, resolves the encoded directory back to a real path, adds --dangerously-skip-permissions / --enable-auto-mode if missing, shows a proof dialog, then cds and execs into Claude. No more "No conversation found" from the wrong folder.

3. The TUI list

cr after a blank paste, or cr 24 / cr --all — see TUI below.

Paste-to-resume parses claude --resume lines and Claude Code session text, so it's Claude Code only. Resuming a Codex session works through the MCP resume_in_terminal tool and the TUI/cards, which emit codex resume <uuid> and cd to the session's recorded cwd.

How paste-to-find works

Finding the right chat from a messy paste isn't one trick — it's a small retrieval pipeline. The index makes it fast; the normalize + score + gate stages make it correct. FTS alone would only get candidates quickly.

The summary/BM25 index (search_sessions) holds only titles + summaries, so pasted transcript text isn't findable there — this is a separate full-text index over normalized session bodies.

Measured results (180-trial calibration + adversarial batch, driven through emux/tmux):

Speed — indexed query (FTS lookup + coverage scoring): p50 6 ms at 166 MB, 94 ms at 3 GB. The first run builds the index (~10 s); after that, incremental refresh re-indexes only changed sessions, so queries stay in the millisecond range. A brute-force scan without an index is ~5.5 s/query and grows linearly with corpus size — the chart at the top is that gap.

TUI

For when your machine died and you just need to get back to work.

pip install resume-resume
resume-resume        # last 4 hours
cr 24                # last 24 hours
cr --all             # everything

Requires Python 3.11+ and Claude Code.

How It Works

1. Scans ~/.claude/projects/ (Claude Code) and ~/.codex/sessions/ (Codex CLI) for JSONL session files

2. Scans all project directories for dirty git state (parallel git status --porcelain)

3. Scores sessions by urgency: session recency (2h half-life) + repo dirty urgency (file count + dirty file recency)

4. Dirty repos bypass age filters — uncommitted work doesn't age out

5. Summarizes via claude -p with Haiku, cached permanently after first run

6. Classifies sessions as human or automated using a gradient boosting model trained on 3,800 sessions — bot sessions hidden by default

7. Surfaces bookmark data (lifecycle badges, next actions, blockers) when present

8. Reports negative space — how many repos were scanned, how many were clean vs dirty

Run /bookmark inside any Claude Code session to capture lifecycle state (done, paused, blocked, handoff) before closing. An auto-bookmark Stop hook captures minimal state when you don't.

Related

• claude-session-commons — Shared session parsing, caching, and classification used by this repo and others

• resume-resume-duet — Web UI companion with session browser and resume-resume:// URL scheme handler

License

MIT