
claude-systems-prompts
All official Claude system prompts in one open-source repo from Claude Fable 5.2 to Claude Haiku 3 23 models from Anthropic's release notes, plus original a…
Install with your AI
Paste into Claude Code, Cursor, or any agent — it reads the repo and wires the tool into your project.
Install and set up claude-systems-prompts (git-clone project) into my current project. Found on https://claudeers.com/claude-systems-prompts Repo: https://github.com/KalarisLabs/claude-systems-prompts Homepage/docs: — Detected install method: git-clone → git clone https://github.com/KalarisLabs/claude-systems-prompts Category: automation. Platforms: cli, api, desktop, web, mobile. Read the repo's README for exact setup and env vars, then install it and wire it into my project. Claudeers Health Verdict: unknown; community-verified: false. Confirm the source before running anything.
git clone https://github.com/KalarisLabs/claude-systems-prompts
// compatibility
| Platforms | cli, api, desktop, web, mobile |
|---|---|
| Operating systems | — |
| AI compatibility | claude |
| License | NOASSERTION |
| Pricing | open-source |
| Language | — |
Claude System Prompts: Every Official Claude System Prompt in One Repo
A dated, exact archive of all official Anthropic system prompts for Claude, from Claude Fable 5.2 back to Claude Haiku 3.

Curated & Extracted by Sayan Chowdhury
All official Claude system prompts for claude.ai, the Claude iOS app, and the Claude Android app, plus original analysis of how Claude is engineered to think, refuse, write, and use tools.
Models at a Glance | Who This Is For | Key Revelations | Full Fable 5 Prompt | Analysis | Highlights
The Prompts Are Public. Almost Nobody Reads Them.
Anthropic publishes Claude's system prompts on its own release notes pages. Twenty-three models worth, hiding in plain sight. And almost nobody reads them.
That should bother you. The system prompt decides what Claude refuses, how it formats your answer, when it searches the web, what it makes of your legal questions, and how it responds to a mental health crisis. It is the most underrated document in AI right now, and it is free to read.
This repo archives all of it. Every official prompt Anthropic ships for claude.ai and the Claude mobile apps, copied exactly, with dates and source links: Claude Fable 5.2 (November 9, 2026) back to Claude Haiku 3 (July 12, 2024). No paraphrase. No summaries. No "reportedly."
What is inside:
models/: all 23 official system prompts, copied exactly, with latest version dates, full version history per model, and direct links to the source page for every releasesystem-prompt/: deep dive on the flagship: the complete ~1,580-line Claude Fable 5 system prompt, tool definitions, and multi-part analysishighlights/: the most surprising findings across the Claude prompt family
Coverage: 23 models. 50 published prompt versions referenced. July 2024 to November 2026. Every file copied exactly from Anthropic's own pages.
Models at a Glance
All 23 official prompts, sorted by family and date. If a prompt is not on this table, it has not been published.
Every archived prompt lives at models/<model>/system-prompt.md. Each file carries its latest version date, lists every version Anthropic has published on that model's page (newest first), and links the canonical source. See the full index with per-model version history in models/README.md.
Note: These are the system prompts for claude.ai and the Claude mobile apps. They do not apply to the Claude API. When Anthropic publishes a newer version of any prompt, this repo is updated to match. See Contributing.
Who This Is For
If you build on, research, dissect, or write about Claude, and you have not read the prompt running behind it, you are shipping on vibes. This archive fixes that.
AI & Machine Learning Engineers
Stop tuning blind. This is the contract your model runs under. Study how Anthropic structures refusal handling, tone control, formatting policy, and tool-use in production, then steal the patterns for your own system prompts. Diff two dated versions of the same model (Claude Fable 5, June 9 vs October 30, 2026) and watch exactly how Anthropic iterates.
AI Safety & Alignment Researchers
The pages everyone quotes from secondary sources, printed word for word: <critical_child_safety_instructions>, <user_wellbeing> crisis protocols, the <anthropic_reminders> classifier system, dual-use capability gating, and evenhandedness guidelines. Fifty dated versions to study drift.
Security Researchers & Red Teamers
You cannot red team what you will not read. The injection defenses, reminder-tag provenance rules, and refusal boundaries are all in here, across every model Anthropic ships for consumers.
Product Builders & Startup Founders
Know what claude.ai gives users out of the box before you rebuild it worse. Memory, artifacts, web search, citations, artifact API calls: the manual is right here.
Students, Educators & AI-Literacy Researchers
Original source documents. Citable, dated, preserved. Teach how frontier assistants are engineered using the real text, not a blog post about it.
Journalists & Tech Writers
Quote Anthropic's own words instead of someone's paraphrase. Trace how safeguards grew from Claude 3 (July 2024) to the Mythos-class Fable 5 line (November 2026).
For the same model family recorded the other way round, off the wire rather than from a live session, see OrcaPromptVault: its 2026-09-02 claude-fable-5-1 captures on Claude Code 2.1.258.18d hold 26,131 characters and 35 tool declarations interactively and 20,806 and 29 under claude -p, as sizes of the scrubbed files, each with the command that reproduces it. The product surface and the chat surface do not get the same prompt, and the archive keeps them apart.
Key Revelations
The Claude Fable 5 archive (extracted from a live claude.ai session) is the clearest window we have into how Anthropic builds its flagship model. Highlights below. Full context in system-prompt/.
Mythos-Class: A New Model Tier Above Opus
Claude Fable 5 belongs to a new "Mythos-class" model tier that sits above Claude Opus in capability. The prompt confirms:
"This iteration of Claude is Claude Fable 5, the first model in Anthropic's new Claude 5 family and part of a new Mythos-class model tier that sits above Claude Opus in capability."
- Claude Fable 5 and Claude Mythos 5 share the same underlying model
- Claude Mythos 5 is available without safety measures to approved organizations only
- Model strings:
claude-fable-5,claude-opus-4-8,claude-sonnet-4-6,claude-haiku-4-5-20251001
The Full Claude Product Ecosystem
The prompt reveals a large product lineup:
| Product | Description |
|---|---|
| Claude Code | Agentic coding tool (CLI, desktop, mobile) |
| Claude Cowork | Agentic knowledge-work desktop app for non-developers |
| Claude in Chrome | Browsing agent |
| Claude in Excel | Spreadsheet agent |
| Claude in PowerPoint | Slides agent |
Claude Cowork can use all of the above as tools, making it a meta-agent that orchestrates other Claude agents.
"Claudeception": API Inside Artifacts
Claude can make Anthropic API calls from within Artifacts, creating AI-powered interactive applications. The prompt uses Sonnet 4 as the internal model:
model: "claude-sonnet-4-20250514" // Always use Sonnet 4
Claude can build applications that call Claude. A recursive AI architecture.
22+ Tools in the Arsenal
The prompt defines an extensive toolset including: web_search, web_fetch, bash_tool, create_file, str_replace, image_search, places_search, places_map_display_v0, recipe_display_v0, ask_user_input_v0, message_compose_v1, suggest_connectors, search_mcp_registry, recommend_claude_apps, fetch_sports_data, weather_fetch, present_files, and more.
Aggressive Copyright Compliance
The prompt contains an extremely detailed copyright enforcement system with hard limits:
- 15+ words from any single source = SEVERE VIOLATION
- ONE quote per source MAXIMUM: after one quote, that source is "CLOSED"
- Default to paraphrasing: quotes should be "rare exceptions"
- Never reproduce song lyrics, poems, or haikus in any form
Self-Harm & Mental Health: Unprecedented Depth
The prompt includes detailed mental health guidelines:
- No substitution techniques that mimic self-harm (ice cubes, rubber bands, sour candy)
- No causal narratives for disordered eating the user hasn't made themselves
- Acknowledge bad experiences with crisis services without totalizing the system
- Never build over-reliance on Claude or push the user to keep talking
Knowledge Cutoff: End of January 2026
Claude's reliable knowledge cutoff is end of January 2026, with the current date set to June 9, 2026.
Every File and Why It Is Here
claude-systems-prompts/
├── README.md ← You are here
├── LICENSE ← MIT License (prompt content © Anthropic, PBC)
├── CONTRIBUTING.md
├── models/ ← ALL official Claude system prompts
│ ├── README.md ← Index of every model, version, and source
│ ├── claude-fable-5-2/system-prompt.md ← exact copy + latest version date
│ ├── claude-fable-5-1/
│ ├── claude-fable-5/
│ ├── claude-opus-5-5/ claude-opus-5/ claude-opus-4-8/ ... claude-opus-3/
│ ├── claude-sonnet-5-5/ claude-sonnet-5/ claude-sonnet-4-6/ ... claude-sonnet-3-5/
│ ├── claude-haiku-5-5/ claude-haiku-4-5/ claude-haiku-3-5/ claude-haiku-3/
│ └── ... ← 23 models, 50 published versions indexed
├── system-prompt/ ← Fable 5 deep dive (original analysis)
│ ├── full-system-prompt.md ← Complete formatted system prompt
│ ├── tool-definitions.md ← All 22+ tool schemas extracted
│ └── analysis/
│ ├── key-findings.md ← Top discoveries and insights
│ ├── model-architecture.md ← What we learn about the model stack
│ ├── behavioral-guidelines.md ← Deep dive into behavior engineering
│ └── fable5-vs-opus-comparison.md ← Mythos-class vs Opus-class comparison
├── highlights/
│ └── surprising-revelations.md ← Most shocking/viral findings
└── assets/
└── claude-system-prompt-graphic.png ← README hero banner
Table of Contents: Fable 5 System Prompt
| Section | Description |
|---|---|
claude_behavior | Core behavioral rules, product info, tone, formatting |
memory_system | Cross-session memory access |
persistent_storage_for_artifacts | Key-value storage API for artifacts |
mcp_app_suggestions | MCP connector/app suggestion system |
computer_use | Skills, file handling, artifact creation |
search_instructions | Web search behaviors and copyright compliance |
using_image_search_tool | Image search guidelines |
Tool Definitions | Complete schemas for all tools |
anthropic_api_in_artifacts | "Claudeception": API-in-artifact system |
citation_instructions | Citation format and rules |
network_configuration | Allowed domains and egress rules |
How to Use This Repo
Browse any model: Start with models/README.md. It indexes every official Claude system prompt by model and version date.
Read the flagship prompt: system-prompt/full-system-prompt.md for the complete, formatted Claude Fable 5 system prompt with section anchors.
Quick highlights: Jump to highlights/surprising-revelations.md for the most interesting findings.
Deep analysis: The system-prompt/analysis/ directory contains breakdowns of the model architecture, behavioral engineering, and a detailed Fable 5 vs Opus 4.8 Comparison.
Why This Matters
System prompts are the invisible architecture behind AI behavior. This archive reveals:
- How Anthropic engineers personality: from warm tone requirements to anti-bullet-point philosophy
- The Mythos tier system: a previously unknown model hierarchy above Opus
- Recursive AI architecture: Claude calling Claude inside Artifacts
- Unprecedented safety nuance: mental health guidelines more detailed than most clinical protocols
- The agentic future: Cowork as a meta-agent orchestrating Claude Code, Chrome, Excel, and PowerPoint
- How safeguards evolve: track
<user_wellbeing>and refusal policy changes across 50 dated versions, from Claude 3 to Fable 5.2
This archive is the primary source document set for understanding how frontier AI models are shaped, and how that shaping changes over time.
The Weird Rules Inside Claude
These are the prompt's own rules, word for word. Anthropic wrote them. Believe them.
- "Claudeception": Fable 5 is explicitly allowed to use the Anthropic API (via Sonnet 4) from within the interactive artifacts it generates.
- The Anti-Bullet-Point Crusade: The prompt aggressively bans Claude from using bullet points or bold text in standard prose, noting it wants to avoid looking like stereotypical "AI slop." It even notes: "Claude never uses bullet points when declining a task; the additional care helps soften the blow."
- No Sycophantic Gratitude: Claude is explicitly told to never thank the user just for reaching out, and to never beg the user to keep talking.
- Military-Grade Copyright: More than 15 words quoted from a single source is defined as a "SEVERE VIOLATION". The prompt includes a 5-step checklist Claude must run before quoting text.
- Clinical Nuance: It forbids suggesting "rubber bands or ice cubes" as self-harm substitutes, noting (like a graduate psychology textbook) that sensory substitutions reinforce neural pathways rather than breaking them.
For the full list of deep dives, see highlights/surprising-revelations.md.
Keywords & Topics
Claude system prompt, Anthropic system prompt, Claude system prompt archive, claude.ai system prompt, Claude Opus system prompt, Claude Sonnet system prompt, Claude Haiku system prompt, Claude Fable 5 system prompt, Claude Fable 5.2, Claude Mythos, Mythos-class model, Anthropic Claude behavioral guidelines, prompt engineering for Claude, AI safety research, frontier model system prompt, Claude API vs claude.ai prompts, open-source AI prompt archive
Community & Contributing
We want to make this repository the premier research hub for Anthropic's Claude models. Found a novel jailbreak, a behavioral quirk, a new prompt version, or want to add an interactive tool?
Please check our Contributing Guidelines and open an Issue using our templates, or submit a PR with your analysis. Community discussion is welcome.
Contributors
Disclaimer
This repository is for educational and research purposes. The archived system prompts are the intellectual property of Anthropic, PBC and are reproduced exactly from Anthropic's official release notes (platform.claude.com). The Claude Fable 5 deep dive was extracted from a publicly accessible claude.ai session. All rights to the original content belong to Anthropic. This repo does not claim ownership of Anthropic's intellectual property.
License
MIT. See LICENSE for details. The archived prompt text itself is Anthropic, PBC intellectual property and is not covered by the MIT grant. See the note in the LICENSE file.
Why Star This Repo
Star this so you hear about it when Anthropic ships a new prompt version, or quietly rewrites the rules behind Claude. That is the whole deal. You also get:
- Real-time updates when Anthropic publishes a new or updated system prompt for any Claude model.
- Community-driven analysis and insights into frontier AI behavior.
- JSON schemas and tool structures you can reverse-engineer for your own AI agents.
If you found this interesting, consider starring the repo!
This repository is curated and maintained by Sayan Chowdhury. All rights to the original system prompt text are owned by Anthropic, PBC.
// faq
What is claude-systems-prompts?
All official Claude system prompts in one open-source repo from Claude Fable 5.2 to Claude Haiku 3 23 models from Anthropic's release notes, plus original analysis, tool schemas, and AI-safety research.. It is open-source on GitHub.
Is claude-systems-prompts free to use?
claude-systems-prompts is open-source under the NOASSERTION license, so it is free to use.
What category does claude-systems-prompts belong to?
claude-systems-prompts is listed under automation in the Claudeers registry of Claude-compatible tools.
// embed badge
[](https://claudeers.com/claude-systems-prompts)
// retro hit counter
[](https://claudeers.com/claude-systems-prompts)
// reviews
// guestbook
// related in Automation & Workflows
The API to search, scrape, and interact with the web at scale. 🔥
🌐 Make websites accessible for AI agents. Automate tasks online with ease.
Taste-Skill - gives your AI good taste. stops the AI from generating boring, generic slop