Extracted system prompts from Anthropic - Claude Fable 5.1, Opus 5, Claude Design, Claude Code. OpenAI - ChatGPT GPT-6-Astra, Codex. Google - Gemini 3.8 Flash, 3.1 Pro, Antigravity. xAI - Grok, Grok Bot, Cursor, Kimi and more! Updated regularly.
This repository collects and archives extracted system prompts from major AI assistants, updated regularly as new model versions ship. For Anthropic products specifically, it includes prompts for Claude Fable 5, Claude Opus 4.8, multiple Claude Code variants (with separate files for the glob tool, grep tool, deferred tools, and bundled skills), Claude Cowork, Claude Design, and integrations such as Claude for Excel, Word, PowerPoint, and iOS. Beyond Anthropic, the repository covers OpenAI's GPT-5.5 Thinking, GPT-5.5 Instant, and Codex; Google's Gemini 3.5 Flash and Antigravity CLI; xAI's Grok; and tools like VS Code Copilot Agent, Cursor, and Perplexity. Where prompts change between versions, the repository links to diff views, such as the documented change from Claude Opus 4.8 to Claude Fable 5. The primary audience is prompt engineers, AI researchers, and developers who want to understand how these systems are instructed, reverse-engineer behavioral constraints, or benchmark prompt changes across model releases. The repository received coverage in The Washington Post in May 2026.
Repository collecting extracted/leaked system prompts from major AI chatbots and coding assistants.
- ✓Open-source license (CC0-1.0)
- ✓Actively maintained (<30d)
- ✓Healthy fork ratio
- ✓Clear description
- ✓Topics declared
- ✓Mature repo (>1y old)
git clone https://github.com/asgeirtj/system_prompts_leaks && cp system_prompts_leaks/*.md ~/.claude/agents/24 items en este repositorio
Deep research harness — fan-out web searches, fetch sources, adversarially verify claims, synthesize a cited report.
Author or improve the run-<unit> skill - a per-project skill that tells agents how to build, launch, and drive this project's app. Use when the user asks to set up the project, get it running, write run instructions, or verify build/run steps work from a clean environment.
Runtime capabilities a published Artifact page can be granted — behavior static HTML cannot provide on its own, such as the page reading live or connected data, remembering what people do on it (a poll, a sign-up sheet, a checklist, a document edited in place — it saves new versions of itself), keeping state shared across viewers, knowing who is viewing, asking Claude a question of its own, storing files people add, or handing the viewer a file to save. Serves this user's live capability roster and the typed call definitions. Load it whenever the user asks for an artifact needing any such runtime behavior.
Diagramming know-how for Artifacts - when a picture earns its place, how to draw one that shows the real mechanism, and the inline-SVG mechanics that keep it legible in both themes.
Research and plan a large-scale change, then execute it in parallel across 5–30 isolated worktree agents that each open a PR.
Automates your Chrome browser to interact with web pages - clicking elements, filling forms, capturing screenshots, reading console logs, and navigating sites. Opens pages in new tabs within your existing Chrome session. Requires site-level permissions before executing (configured in the extension).
Review the current diff, or a PR number/branch/path target, for correctness bugs and reuse/simplification/efficiency cleanups at the given effort level (low/medium: fewer, high-confidence findings; high→max: broader coverage, may include uncertain findings; ultra: deep multi-agent review in the cloud); with no level given, it reuses the level you typed last. Pass --comment to post findings as inline PR comments, or --fix to apply the findings to the working tree after the review. For ultra on a GitHub.com PR target, --post asks to post the finished review’s findings to the PR as a single comment from the user’s GitHub account (not a review; the launch dialog still confirms in interactive sessions, while non-interactive mode posts on the flag alone) and --no-post hides that option.
Push a React design system to claude.ai/design. This runs a converter that bundles the real component code (from Storybook or a bare package) and uploads it. Use when the user runs /design-sync or says "sync my design system to Claude Design".
Create a design canvas - a multi-artboard visual design published as an Artifact that runs Claude Design's canvas editor (an early preview of Claude Design inside Claude Code). You DRAFT the design as .dc.html artboards laid out on one pan/zoom canvas; where saving is enabled for the user's account they refine every element visually (click-to-select, a properties panel, inline text editing, undo/redo) and Save publishes a new version for everyone, otherwise they get a view-and-export (PNG/PDF) preview of your draft. Good for UI mockups and screen flows, landing pages, marketing and social graphics, and print pieces - posters, flyers, brochures as single-page artboards; memos and reports as one flowing artboard. Use when someone wants a design, mockup, wireframe, UI or screen design, landing page, poster, flyer, brochure, banner, card, one-pager, or any visual layout they would rather tweak by hand than in code. Only for CREATING or re-seeding a canvas; an existing one is edited in its published Artifact.
Health-check the user's Claude Code setup and fix issues: diagnose installation health — what the `claude doctor` terminal diagnostics cover — from local data (duplicate or leftover installs, PATH, unparseable settings files, broken or colliding agent definitions, skills whose frontmatter fails to parse); find unused skills, MCP servers, and plugins versus their context cost and disable dead weight; deduplicate local CLAUDE.md files against checked-in ones; trim checked-in CLAUDE.md files by cutting content a session could derive from the codebase (directory layouts, tech-stack lists, architecture overviews) while keeping gotchas, rationale, and non-standard conventions; migrate always-loaded CLAUDE.md guidance into lazy skills and nested CLAUDE.md files; flag slow hooks and context-heavy extensions; check the installed version is current; make auto mode the default permission mode; and pre-approve frequently denied read-only commands. Use when the user asks for a doctor run, checkup, audit, tune-up, or cleanup of their Claude Code setup or configuration.
Scan your transcripts for common read-only Bash and MCP tool calls, then add a prioritized allowlist to project .claude/settings.json to reduce permission prompts.
Use when the user wants to customize keyboard shortcuts, rebind keys, add chord bindings, or modify ~/.claude/keybindings.json. Examples: "rebind ctrl+s", "add a chord shortcut", "change the submit key", "customize keybindings".
Run a prompt or slash command on a recurring interval (e.g. /loop 5m /foo). Omit the interval to let the model self-pace.
Launch and drive this project's app to see a change working. Use when asked to run, start, or screenshot the app, or to confirm a change works in the real app (not just tests). First looks for a project skill that already covers launching the app; otherwise falls back to built-in patterns per project type (CLI, server, TUI, Electron, browser-driven, library).
Create, update, list, or run scheduled cloud agents (routines) that execute on a cron schedule.
Review the changed code for reuse, simplification, efficiency, and altitude cleanups, then apply the fixes. Quality only — it does not hunt for bugs; use /code-review for that.
Use this skill to configure the Claude Code harness via settings.json. Automated behaviors ("from now on when X", "each time X", "whenever X", "before/after X") require hooks configured in settings.json - the harness executes these, not Claude, so memory/preferences cannot fulfill them. Also use for: permissions ("allow X", "add permission", "move permission to"), env vars ("set X=Y"), hook troubleshooting, or any changes to settings.json/settings.local.json files. Examples: "allow npm commands", "add bq permission to global settings", "move permission to user settings", "set DEBUG=true", "when claude stops show X". For simple settings like theme/model, suggest the /config command.
Resumen de Subagents
<p align="center">
<sub>Thanks to</sub>
</p>
<p align="center">
<a href="https://go.asgeirtj.workers.dev/latitude">
<img src="assets/latitude-dark.png" alt="Latitude Logo" width="700"/>
</a>
</p>
<div align="center" markdown="1">
### [Open Source Agent Analytics](https://go.asgeirtj.workers.dev/latitude)
[Understand your agents in production](https://go.asgeirtj.workers.dev/latitude)
</div>
---
> **The Washington Post** built its interactive story on prompts from this repo: [See the hidden rules behind AI. Then use them to rewrite this article.](https://archive.today/pYtPk) (May 11, 2026)
>
> **CEPS' AI World** built a live data dashboard from this repo's files: [System prompts and what they tell us about the chat before the chat](https://aiworld.eu/story/system-prompts-and-what-they-tell-us-about-the-chat-before-the-chat) (July 10, 2026)
# System Prompts Leaks
Leaked system prompts, captured verbatim — the hidden instructions and rules that ChatGPT, Claude, Gemini, Grok and every other AI chatbot receives before your first message.
[](http://makeapullrequest.com)
## Most recent additions/changes
| What | Date | Link |
|------|------|------|
| **Claude Code headless (Fable 5.1)** | September 5, 2026 | [Claude Code headless system prompt (Fable 5.1)](Anthropic/claude-code/claude-code-headless-fable-5.1.md) |
| **Codex GPT-6-Astra** | September 4, 2026 | [Codex GPT-6-Astra system prompt](OpenAI/Codex/gpt-6-astra.md) |
| **Claude Fable 5.1** | September 1, 2026 | [Claude Fable 5.1 system prompt](Anthropic/claude-fable-5.1.md) · [Claude Code (Fable 5.1)](Anthropic/claude-code/claude-code-fable-5.1.md) |
| **Grok 4.6** | August 29, 2026 | [Grok 4.6 system prompt](xAI/grok-4.6.md) |
| **Grok Bot** | August 21, 2026 | [Grok Bot System Prompt](xAI/grok-bot.md) |
| **Gemini 3.7 Flash** | August 18, 2026 | [Gemini 3.7 Flash System Prompt](Google/gemini-3.7-flash.md) |
| **Muse Code (Meta's coding CLI)** | August 17, 2026 | [Muse Code system prompt](Meta/muse-code.md) |
| **Claude Cowork (new capture + setup skills)** | August 17, 2026 | [Claude Cowork system prompt](Anthropic/claude-cowork/claude-cowork.md) · [setup skills](Anthropic/claude-cowork) |
| **Claude Science** | August 14, 2026 | [Claude Science system prompt](Anthropic/claude-science.md) |
| **Codex GPT-5.6 (Sol variant)** | July 26, 2026 | [Codex GPT-5.6 system prompt](OpenAI/Codex/gpt-5.6.md) |
| **Grok 4.5** | July 26, 2026 | [Grok 4.5 system prompt](xAI/grok-4.5.md) |
| **Claude Opus 5** | July 24, 2026 | [Claude Opus 5 system prompt](Anthropic/claude-opus-5.md) · [Claude Code (Opus 5)](Anthropic/claude-code/claude-code-opus-5.md) |
| **Claude Design (full prompt + 53 tools + 22 skills + 10 starter components)** | July 23, 2026 | [Claude Design system prompt](Anthropic/claude-design/claude-design.md) · [skills](Anthropic/claude-design/skills) · [starter components](Anthropic/claude-design/starter-components) |
| **Perplexity** | July 17, 2026 | [Perplexity AI system prompt](Perplexity/perplexity-ai.md) |
| **Claude Code (new models)** | July 16, 2026 | [Claude Code system prompt (Fable 5)](Anthropic/claude-code/claude-code-fable-5.md) · [Sonnet 5](Anthropic/claude-code/claude-code-sonnet-5.md) |
| **OpenCode · Pi · CommandCode** | July 16, 2026 | [OpenCode system prompt](OpenCode/opencode.md) · [Pi system prompt](Pi/instructions.md) · [CommandCode CLI system prompt](Misc/commandcode-cli.md) |
| **Kimi K2.6** | July 14, 2026 | [Kimi K2.6 system prompt](Kimi/kimi-2.6.md) |
| **Perplexity Deep Research** | July 14, 2026 | [Perplexity Deep Research system prompt](Perplexity/deep-research.md) |
| **ChatGPT 5.6** | July 10, 2026 | [ChatGPT 5.6 system prompt (Sol, extra high)](OpenAI/gpt-5.6-sol.md) · [Codex GPT-5.6 system prompt](OpenAI/Codex/gpt-5.6.md) |
---
## Anthropic — Claude system prompts
### Claude.ai system prompts (web, desktop & mobile)
| Model | Prompt |
|-------|--------|
| **Claude Fable 5.1** | [**Claude Fable 5.1 system prompt**](Anthropic/claude-fable-5.1.md) |
| Claude Fable 5 | [Claude Fable 5 system prompt](Anthropic/claude-fable-5.md) |
| **Claude Opus 5** | [**Claude Opus 5 system prompt**](Anthropic/claude-opus-5.md) |
| Claude Opus 4.8 | [Claude Opus 4.8 system prompt](Anthropic/claude-opus-4.8.md) |
| **Claude Sonnet 5** | [**Claude Sonnet 5 system prompt**](Anthropic/claude-sonnet-5.md) |
| Claude Opus 4.7 | [Claude Opus 4.7 system prompt](Anthropic/claude-opus-4.7.md) |
| Claude Opus 4.6 | [Claude Opus 4.6 system prompt](Anthropic/claude-opus-4.6.md) · [No tools](Anthropic/claude-opus-4.6-no-tools.md) |
| Claude Sonnet 4.6 | [Claude Sonnet 4.6 system prompt](Anthropic/claude-sonnet-4.6.md) · [No tools](Anthropic/claude-sonnet-4.6-no-tools.md) |
| Injected reminders | [Claude.ai injected reminders](Anthropic/anthropic_reminders.md) |
### Claude Code system prompts
| Component | Prompt |
|-----------|--------|
| **Claude Code (Fable 5.1)** | [**Claude Code system prompt (Fable 5.1)**](Anthropic/claude-code/claude-code-fable-5.1.md) |
| Claude Code (Fable 5) | [Claude Code system prompt (Fable 5)](Anthropic/claude-code/claude-code-fable-5.md) |
| **Claude Code (Opus 5)** | [**Claude Code system prompt (Opus 5)**](Anthropic/claude-code/claude-code-opus-5.md) |
| Claude Code (Opus 4.8) | [Claude Code system prompt (Opus 4.8)](Anthropic/claude-code/claude-code-opus-4.8.md) |
| **Claude Code (Sonnet 5)** | [Claude Code system prompt (Sonnet 5)](Anthropic/claude-code/claude-code-sonnet-5.md) |
| Claude Code (older models) | [Opus 4.7](Anthropic/claude-code/claude-code-opus-4.7.md) · [Opus 4.6](Anthropic/claude-code/claude-code-opus-4.6.md) · [Sonnet 4.6](Anthropic/claude-code/claude-code-sonnet-4.6.md) · [Haiku 4.5](Anthropic/claude-code/claude-code-haiku-4.5.md) |
| Subagents | [Claude Code subagent system prompts](Anthropic/claude-code/agents) |
| Skills & commands | [Claude Code skills](Anthropic/claude-code/skills) · [Slash commands](Anthropic/claude-code/commands) |
| MCP servers | [Claude in Chrome MCP server system prompt](Anthropic/claude-in-chrome.md) |
| Docs assistant | [docs.claude.com assistant instructions](Anthropic/claude-code/claude-code-docs-assistant.md) |
### Claude integrations
| Product | Prompt |
|---------|--------|
| **Claude Design** | [**Claude Design system prompt**](Anthropic/claude-design/claude-design.md) · [skills](Anthropic/claude-design/skills) · [starter components](Anthropic/claude-design/starter-components) |
| **Claude Cowork** | [Claude Cowork system prompt](Anthropic/claude-cowork/claude-cowork.md) · [Dispatch](Anthropic/claude-cowork/claude-cowork-dispatch.md) · [setup skills](Anthropic/claude-cowork) |
| **Claude Science** | [Claude Science system prompt](Anthropic/claude-science.md) |
| Claude for Microsoft 365 | [Claude for Excel](Anthropic/claude-for-excel.md) · [Claude for Word](Anthropic/claude-for-word.md) · [Claude in PowerPoint](Anthropic/claude-in-powerpoint.md) |
| Claude in Chrome | [Claude in Chrome extension system prompt](Anthropic/claude-in-chrome.md) |
| Claude iOS app | [Claude mobile iOS system prompt](Anthropic/claude-mobile-ios.md) |
## OpenAI — ChatGPT system prompts
| Model | Prompt |
|-------|--------|
| **ChatGPT 5.6 Sol** | [**ChatGPT 5.6 system prompt (Sol, extra high)**](OpenAI/gpt-5.6-sol.md) |
| **ChatGPT 5.5 Thinking** | [**ChatGPT 5.5 Thinking system prompt**](OpenAI/gpt-5.5-thinking.md) |
| **ChatGPT 5.5 Instant** | [**ChatGPT 5.5 Instant system prompt**](OpenAI/gpt-5.5-instant.md) |
| ChatGPT 5.4 | [ChatGPT 5.4 Thinking system prompt](OpenAI/gpt-5.4-thinking.md) |
| ChatGPT 5.3 | [ChatGPT 5.3 Instant system prompt](OpenAI/gpt-5.3-instant.md) |
| ChatGPT 5.2 | [ChatGPT 5.2 Thinking system prompt](OpenAI/gpt-5.2-thinking.md) |
| ChatGPT 5 | [ChatGPT 5 Thinking system prompt](OpenAI/gpt-5-thinking.md) · [Agent mode](OpenAI/chatgpt-gpt-5-agent-mode.md) |
| **ChatGPT Atlas** | [ChatGPT Atlas system prompt](OpenAI/chatgpt-atlas.md) |
| ChatGPT 4.5 | [ChatGPT 4.5 system prompt](OpenAI/chatgpt-4.5.md) |
| ChatGPT 4o | [ChatGPT 4o system prompt](OpenAI/gpt-4o.md) · [Deprecation preparedness](OpenAI/chatgpt-4o-deprecation-preparedness-prompt.md) |
| Voice modes | [ChatGPT advanced voice mode system prompt](OpenAI/gpt-4o-advanced-voice-mode.md) · [Legacy voice mode](OpenAI/gpt-4o-legacy-voice-mode.md) |
| Personalities | [ChatGPT personality instructions](OpenAI/chatgpt-personality-instructions.md) |
| Memory | [ChatGPT advanced memory system prompt](OpenAI/tool-advanced-memory.md) |
### Codex system prompts
| Model | Prompt |
|-------|--------|
| **Codex GPT-5.6** | [**Codex GPT-5.6 system prompt (Terra/Luna)**](OpenAI/Codex/gpt-5.6.md) · [Sol](OpenAI/Codex/gpt-5.6-sol.md) |
| **Codex GPT-5.5** | [Codex GPT-5.5 system prompt](OpenAI/Codex/gpt-5.5.md) · [Full prompt](OpenAI/Codex/codex-full.md) · [Friendly](OpenAI/Codex/personality_friendly_gpt-5.5.md) · [Pragmatic](OpenAI/Codex/personality_pragmatic_gpt-5.5.md) |
| Codex GPT-5.4 | [Codex GPT-5.4 system prompt](OpenAI/Codex/gpt-5.4.md) · [Mini](OpenAI/Codex/gpt-5.4-mini.md) |
| Codex Spark | [Codex Spark system prompt](OpenAI/Codex/gpt-5.3-codex-spark.md) |
| Codex modes | [Plan mode](OpenAI/Codex/plan_mode.md) · [Auto-review](OpenAI/Codex/codex-auto-review.md) · [Computer use](OpenAI/Codex/computer-use.md) · [Control Chrome](OpenAI/Codex/control-chrome.md) · [In-app browser](OpenAI/Codex/control-in-app-browser.md) |
| Personas | [Friendly](OpenAI/Codex/personality_friendly.md) · [Pragmatic](OpenAI/Codex/personality_pragmatic.md) |
### API-injected prompts
| Model | Prompt |
|-------|--------|
| Lo que la gente pregunta sobre system_prompts_leaks
¿Qué es asgeirtj/system_prompts_leaks?
+
asgeirtj/system_prompts_leaks es subagents para el ecosistema de Claude AI. Extracted system prompts from Anthropic - Claude Fable 5.1, Opus 5, Claude Design, Claude Code. OpenAI - ChatGPT GPT-6-Astra, Codex. Google - Gemini 3.8 Flash, 3.1 Pro, Antigravity. xAI - Grok, Grok Bot, Cursor, Kimi and more! Updated regularly. Tiene 64.7k estrellas en GitHub y su última actualización registrada es del 2026-09-09.
¿Cómo se instala system_prompts_leaks?
+
Puedes instalar system_prompts_leaks clonando el repositorio (https://github.com/asgeirtj/system_prompts_leaks) o siguiendo las instrucciones del README en GitHub. ClaudeWave también te ofrece bloques de instalación rápida en esta misma página.
¿Es seguro usar asgeirtj/system_prompts_leaks?
+
Nuestro agente de seguridad ha analizado asgeirtj/system_prompts_leaks y le ha asignado un Trust Score de 100/100 (tier: Verified). Revisa el desglose completo de comprobaciones superadas y flags en esta página.
¿Quién mantiene asgeirtj/system_prompts_leaks?
+
asgeirtj/system_prompts_leaks es mantenido por asgeirtj. La última actividad registrada en GitHub es del 2026-09-09, con 53 issues abiertos.
¿Hay alternativas a system_prompts_leaks?
+
Sí. En ClaudeWave puedes explorar subagents similares en /categories/agents, ordenados por popularidad o actividad reciente.
Despliega system_prompts_leaks en tu cloud
Lleva este repo a producción en minutos. Cada plataforma genera su propio entorno con variables de entorno editables.
¿Mantienes este repo? Añade un badge a tu README
Pega el badge en tu README de GitHub para mostrar que está auditado por ClaudeWave. Cada badge enlaza de vuelta a esta página y muestra el Trust Score actual.
[](https://claudewave.com/repo/asgeirtj-system-prompts-leaks)<a href="https://claudewave.com/repo/asgeirtj-system-prompts-leaks"><img src="https://claudewave.com/api/badge/asgeirtj-system-prompts-leaks" alt="Featured on ClaudeWave: asgeirtj/system_prompts_leaks" width="320" height="64" /></a>Más Subagents
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
The agent that grows with you
Java 面试 & 后端通用面试指南，覆盖计算机基础、数据库、分布式、高并发、系统设计与 AI 应用开发
Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without rebuilding the stack.
The agent engineering platform.
Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.