SOMEONE LEAKED THE FULL SYSTEM PROMPT OF CLAUDE FABLE 5 — AND IT'S SITTING ON GITHUB WITH 26,000 STARS.
Anthropic launched Fable 5 on June 9.
The prompt was public within 24 hours.
120,000 characters.
1,585 lines.
27,000 tokens.
Every hidden instruction exposed.
CL4R1T4S repo by Pliny the Liberator.
The same repo that previously leaked ChatGPT, Gemini, Grok, Cursor, Lovable, Replit, and Perplexity.
Here is what it actually reveals:
Fable 5 and Mythos 5 share the same underlying model. Fable gets extra safety layers. Mythos doesn't, but is restricted to approved organizations only.
Knowledge cutoff is end of January 2026. Not May. Not March. January.
Claude is explicitly instructed to avoid bullet points, headers, and lists unless the user asks. That question everyone had about why Claude writes in prose now is answered in line 847.
Copyright hard limit: quoting 15 or more words from any single source is flagged as a severe violation. One quote per source maximum. After one quote, that source is closed.
A new persistent storage API lets artifacts store and retrieve data across sessions using key-value pairs.
MCP app connectors are built directly into the prompt. Claude can search for and suggest third-party integrations mid-conversation.
The prompt includes detailed instructions for Claude Cowork, Claude in Chrome, Claude in Excel, and Claude in PowerPoint.
This prompt is specific to
claude.ai's consumer interface. API users get no system prompt. Claude Code has its own separate instructions entirely.
700,000 views in the first two days.
26,400 GitHub stars.
4,700 forks.