🪨 why use many token when few token do trick — Claude Code skill that cuts 65% of tokens by talking like caveman
CLI search for coding agent sessions
fast-resume indexes local conversation history from supported coding agents and lets you search across titles, prompts, responses, folders, and dates. It uses a terminal UI for browsing and a stable JSON mode for agent use, so you can find a session and resume it without digging through logs.
Builders who want to search and resume past sessions in Claude Code, Codex, Cursor, and similar agents.
You can find the session you want and jump back into it instead of re-reading old history.
What it does
Full-text session search
Searches titles, directories, user messages, and assistant replies across supported agent histories.
Typo-tolerant ranking
Uses Tantivy search to find close matches and rank exact hits higher.
Filters in the CLI and TUI
Lets you narrow results by agent, directory, and date from the command line or inside the search box.
Session preview and resume handoff
Shows conversation previews, copies resume commands, or hands off directly to the original agent.
Background refresh
Starts from the current index and refreshes changed sessions in the background.
Agent-friendly JSON output
Prints stable JSON for tools that want to consume results without parsing the table.
Bundled agent skill
Includes an Agent Skill file with search, pagination, and resume guidance for agents.
How to get it
- 1Homebrew is the simplest option on macOS and Linux
brew tap angristan/tap brew install fast-resume
- 2Nix users can run directly from the source package
nix run github:angristan/fast-resume
- 3You can also install with npm or a binary wheel with uv
npm install --global fast-resume # or uv tool install fast-resume
- 4Coding agents can use the stable JSON output instead of parsing the human table
fr --json --limit 10 "dir:backend authentication bug" fr --json --limit 10 --offset 10 "dir:backend authentication bug"
README
fast-resume
Search and resume conversations across Claude Code, Codex, Pi, and more, all from one terminal UI.
Coding agents can resume previous sessions, but searching those sessions is often limited or title-only. fast-resume builds a single full-text index over your local agent history so you can find a prompt, response, project, or session and jump straight back in.
https://github.com/user-attachments/assets/60f6e128-2ae4-431d-8a87-097f600b6d04
Highlights
- Search titles, directories, user messages, and assistant responses across every supported agent.
- Find imperfect matches with typo-tolerant Tantivy search and exact-match ranking.
- Filter by agent, directory, and date from the command line or directly in the search box.
- Preview conversations, copy resume commands, or hand off directly to the original agent.
- Start immediately from the existing index while changed sessions refresh in the background.
- Use agent artwork, mouse controls, responsive filters, and compact layouts in the Ratatui TUI.
Supported agents
Antigravity CLI, Claude Code, Codex, Copilot CLI, Copilot in VS Code, Crush, Cursor CLI, Grok Build, Kimi Code, OpenCode, Pi, and Vibe. See how it works for storage formats and resume behavior.
Install
Homebrew is the simplest option on macOS and Linux:
brew tap angristan/tap
brew install fast-resume
Nix users can run directly from the source package:
nix run github:angristan/fast-resume
You can also install with npm or a binary wheel with uv:
npm install --global fast-resume
# or
uv tool install fast-resume
See the installation guide for Nix, npm, uvx, Cargo, supported platforms, and terminal recommendations.
Quick start
# Search all sessions interactively
fr
# Start with a query or filters
fr "authentication bug"
fr -a codex -d backend "api error"
# Search without opening the TUI
fr --list "agent:claude date:<2d auth"
# Inspect or rebuild the local index
fr --stats
fr --rebuild
Inside the TUI, use the arrow keys to select a session and press Enter to resume it. Tab completes filters or cycles agents, Ctrl+P toggles the preview, and Ctrl+Y copies the resume command.
Agent usage
Coding agents can use the stable JSON output instead of parsing the human table:
fr --json --limit 10 "dir:backend authentication bug"
fr --json --limit 10 --offset 10 "dir:backend authentication bug"
Run fr --agent-context to print the bundled Agent Skill with safe search, pagination, and resume guidance.
Documentation
- Installation — packages, platforms, terminals, and upgrades
- Usage — search syntax, CLI options, keybindings, yolo mode, and statistics
- How it works — adapters, indexing, search, refresh safety, and resume handoff
- Development — local setup, validation, project layout, and release packaging
Configuration
No configuration is required. The Tantivy index follows the XDG Base Directory specification. It lives at $XDG_CACHE_HOME/fast-resume/tantivy_index when XDG_CACHE_HOME is an absolute path, or at ~/.cache/fast-resume/tantivy_index by default. It is rebuilt automatically when its schema changes.
To reset it manually:
rm -rf "${XDG_CACHE_HOME:-$HOME/.cache}/fast-resume"
fr --rebuild
License
MIT
Files in the repo
- .github
- assets
- docs
- nix
- npm
- python
- skills
- src
- tests
- .gitignore
- .pre-commit-config.yaml
- .releaserc.json
- AGENTS.md
- Cargo.lock
- Cargo.toml
- CHANGELOG.md
- commitlint.config.mjs
- flake.lock
- flake.nix
- LICENSE
- pyproject.toml
- README.md
Discussion (0)
Ask about usage, or say what you built with itSign in to join the discussion.
No comments yet. Be the first to say what this is good for.
More tools
The best-benchmarked open-source AI memory system. And it's free.
Orca is the ADE for working with a fleet of parallel agents. Run any coding agent with your own subscription. Available on desktop, mobile and remote runtime.

A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io
Never stop coding. Free MIT AI gateway: one endpoint, 352 providers (150+ free), 1200+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback, RTK+Caveman compression saves 15-95% tokens, MCP/A2A, Desktop/PWA. Built by 550+ contributors
Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.