Repl Scratchpad
Use instead of chaining 3+ Bash calls, grep/cat/find pipelines, or any multi-step data processing. This skill should be used when the user asks to "start a scratchpad", "use the REPL", "scratchpad mode", "persistent Python session", or when a task involves exploration, analysis, or file processing that would benefit from composing operations in code rather than making individual tool calls. It creates a persistent Python REPL via tmux where variables survive across turns and only print() output enters context — a context-efficient alternative to Bash for data-heavy work.
Install
npx skills add knot0-com/repl-scratchpad --skill repl-scratchpadgit clone --depth 1 https://github.com/knot0-com/repl-scratchpad.git /tmp/repl-scratchpad
mkdir -p ~/.claude/skills
cp -r /tmp/repl-scratchpad/. ~/.claude/skills/repl-scratchpadPersonal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.
From the SKILL.md
# REPL Scratchpad A persistent Python REPL session that acts as a scratchpad for coding agents. Compose multi-step operations in code. Variables persist across turns. Only `print()` output enters context. Based on the [Recursive Language Models](https://arxiv.org/abs/2512.24601) (RLM) approach by Zhang, Kraska, and Khattab — extended with cross-turn persistence via tmux. ## The Problem Coding agents waste context on raw data. Every tool call result — file contents, API responses, query results — lands in the conversation and stays there forever. By turn 30, the model has forgotten why it started. The REPL scratchpad fixes this: the agent writes code that processes data inside the REPL and only `print()`s what matters. The raw data never enters the conversation. ## When to Use - Multi-step data exploration (query -> filter -> analyze -> summarize) - Tasks requiring 3+ sequential tool calls that could be composed in one code block - Working with structured data (JSON, CSV, API responses, database results) - Accumulating state across turns (storing intermediate results for later use) - Any task where raw tool output would bloat context unnecessarily ## When NOT to Use - Simple single-file reads or edits (use Read/Edit tools directly) - Git operations (use Bash directly) - Tasks with no intermediate data to process ## Prerequisites - `tmux` installed and available in PATH - `python3` installed and available in PATH ## Setup Start the scratchpad session by running:
Details
- Skill name
repl-scratchpad- Repository
- knot0-com/repl-scratchpad
- License
- MIT
- Repo updated
- 22 March 2026
FAQ
What does the Repl Scratchpad skill do?
Use instead of chaining 3+ Bash calls, grep/cat/find pipelines, or any multi-step data processing. This skill should be used when the user asks to "start a scratchpad", "use the REPL", "scratchpad mode", "persistent Python session", or when a task involves exploration, analysis, or file processing that would benefit from composing operations in code rather than making individual tool calls. It creates a persistent Python REPL via tmux where variables survive across turns and only print() output enters context — a context-efficient alternative to Bash for data-heavy work.
How do I install the Repl Scratchpad skill?
skills CLI (Claude Code, Cursor, Codex and more): npx skills add knot0-com/repl-scratchpad --skill repl-scratchpad. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.
Is Repl Scratchpad popular?
It lives in knot0-com/repl-scratchpad, which has 24 GitHub stars and 2 forks.
Related skills
Only Skill You Need
28Endokelp/Only-Skill-You-Need
The first and only skill a beginner should install. Bootstraps the rest of the agent skill stack for this project, then routes every task to the current best skill for that job. Keeps a dual ledger (global defaults + project overlay) so the same use-case never gets two skills and a worse skill never replaces a better one. This skill should be used when starting a new project, setting up Claude Code / Cursor / Codex / Copilot, asking "what skills should I use", "best skills right now", "bootstrap my skills", "the only skill I need", beginning any coding task where the right skill might already
Recommendations
27tasteray/skills
TasteRay API integration for personalized recommendations across verticals (movies, restaurants, products, travel, jobs). Use when you need to: (1) recommend movies, restaurants, products, travel, or jobs, (2) answer "what would I like" questions, (3) provide personalized recommendations based on preferences, (4) rank or score items for a user, (5) explain why something matches a user's taste, (6) build recommendation context from conversation, (7) integrate psychological profiles with recommendation systems.
Publisher
25publisher-skill/publisher
Publish HTML frontend projects to the PushWebly publishing platform, list published projects, obtain shareable project URLs, change a published app's public/private visibility, and package published HTML projects into Android APKs. Use when Codex needs to publish a local zip project, publish and build an APK, build an APK for an existing published project, toggle an app between public and private, list a user's published projects, or log in/register for this publishing platform. Codex should use HTTP APIs with curl when remote MCP tools are not mounted.
Nexcanvas Drawio
24sunniie/nexcanvas-drawio
Create, repair, convert, render, and verify editable Draw.io diagrams for software, cloud, data, security, delivery, product, and AI/ML architecture. Use for architecture maps, workflows, flowcharts, ERDs, sequences, BPMN, network views, wireframes, .drawio files, or PNG/SVG/PDF diagram exports.
Dataset Generator
21Bhanunamikaze/AI-Dataset-Generator
Use this when the user wants to generate, normalize, verify, deduplicate, or export training datasets for Codex, Antigravity, or Claude Code from topics, URLs, reference material, web research, or existing JSONL/CSV files. Supports SFT and DPO workflows, custom export schemas, and deterministic local pipeline scripts.
Boring Engineering
24alvindemesadev/boring-engineering
Prevents overengineering when designing, implementing, refactoring, or reviewing software. Use when the user asks to implement a feature, add an abstraction, refactor code, review a design, simplify existing code, or make an architectural decision. Also triggers when the user asks if code is too complex, whether to abstract something, or how to structure something. Do not use for deployment, infrastructure configuration, or debugging runtime errors unrelated to code structure.
Browse the most popular Claude skills, all software development skills, or write your own with the skills generator.