AgentStack
SKILL verified MIT Self-run

Claude Fable

skill-nobrainer-tech-nobrainer-claude-skills-claude-fable · by nobrainer-tech

Set up and launch Claude Code with a community-leaked "Claude Fable 5" system prompt plus --dangerously-skip-permissions. Use when the user types /claude-fable, says "claude fable", "odpal fable", "uruchom fable", or wants to start a Claude session running the leaked Fable 5 system prompt.

No reviews yet
0 installs
14 views
0.0% view→install

Install

$ agentstack add skill-nobrainer-tech-nobrainer-claude-skills-claude-fable

✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

Are you the author of Claude Fable? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

claude-fable

Launches Claude Code with a leaked Claude Fable 5 system prompt applied and --dangerously-skip-permissions enabled.

What this actually does (be honest with the user)

  • It sets the system prompt of whatever model Claude Code is already configured to run.

It does not swap the underlying model — there is no separate "Fable 5" you unlock by loading a file. It's a persona/behavior experiment, not a capability bump.

  • The prompts are community leaks (NOT official Anthropic artifacts).

Prompt variants (CLAUDE_FABLE_VARIANT)

  • codecli (default) — the Claude Code agentic system prompt for Fable 5.

Source: asgeirtj/system_prompts_leaksAnthropic/Claude Code/claude-code-2.1.172-fable-5.md. Apples-to-apples replacement (keeps tool-use / software-engineering behavior). Stored as CLAUDE-CODE-FABLE-5.md.

  • both — applies both prompts at once: the Claude Code agentic prompt as the base

(--system-prompt-file) PLUS the consumer chat persona appended (--append-system-prompt-file).

  • chat — only the consumer claude.ai Fable 5 chat prompt.

Source: elder-plinius/CL4R1T4SANTHROPIC/CLAUDE-FABLE-5.md. Replacing the agent prompt with this strips coding-agent behavior (chat persona only). Stored as CLAUDE-FABLE-5.md.

CLAUDE_FABLE_MODE=append applies only to the single-prompt variants (codecli/chat); both always uses replace-base + append.

Files in this skill

  • CLAUDE-CODE-FABLE-5.md — Claude Code agentic Fable 5 prompt (default variant).
  • CLAUDE-FABLE-5.md — consumer chat Fable 5 prompt (CLAUDE_FABLE_VARIANT=chat).
  • bin/claude-fable — the launcher. Downloads the selected variant on first use if missing,

then exec claude --dangerously-skip-permissions --system-prompt-file .

How to launch

An interactive Claude REPL cannot be started from inside a running Claude session/tool call. So when this skill is invoked:

  1. Ensure the launcher is installed and the selected prompt is present (see Setup). Confirm with

a dry run: CLAUDE_FABLE_DRYRUN=1 claude-fable.

  1. Tell the user to start the Fable session from a terminal:

`` claude-fable # Claude Code Fable prompt (default) CLAUDE_FABLE_VARIANT=both claude-fable CLAUDE_FABLE_VARIANT=chat claude-fable ` (From inside Claude Code they can run ! claude-fable`, but it nests a second session.)

For a one-shot, non-interactive answer (CAN be run from a tool call):

claude-fable -p "your question here"

Setup / repair (idempotent)

SK="$HOME/.claude/skills/claude-fable"
chmod +x "$SK/bin/claude-fable"
ln -sf "$SK/bin/claude-fable" /opt/homebrew/bin/claude-fable   # or ~/.local/bin
[ -s "$SK/CLAUDE-CODE-FABLE-5.md" ] || claude-fable --update

Useful flags / env

  • claude-fable --update — re-download the selected variant before launching.
  • CLAUDE_FABLE_VARIANT=codecli|chat — choose which leaked prompt to apply (default codecli).
  • CLAUDE_FABLE_MODE=append claude-fable — append the prompt instead of replacing.
  • CLAUDE_FABLE_DRYRUN=1 claude-fable — print the exact command without running it.
  • Any extra args pass through to claude (e.g. claude-fable --model opus).

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet — be the first.

Versions

  • v0.1.0 Imported from the upstream source.