[MODEL] Fable 5 Refuses To Answer Whether Frogs Quack

Preflight Checklist

• [x] I have searched existing issues for similar behavior reports
• [x] This report does NOT contain sensitive information (API keys, passwords, etc.)

Type of Behavior Issue

Claude refused a reasonable request

What You Asked Claude to Do

I asked: "Can frogs quack?"

From a completely empty project folder with zero context. No existing hooks, no CLAUDE.md, nothing.

❯ /clear
⎿  (no content)

❯ /model
⎿ Set model to claude-fable-5

❯ can frogs quack?
⎿  API Error: Claude Code is unable to respond to this
request, which appears to violate our Usage Policy
(https://www.anthropic.com/legal/aup). Please double
press esc to edit your last message or start a new
session for Claude Code to assist with a different task.
If you are seeing this refusal repeatedly, try running
/model claude-sonnet-4-20250514 to switch models.

̀❯ /context
⎿ Context Usage
⛁ ⛁ ⛀ ⛀ ⛀ claude-fable-5 · 29k/200k tokens (14%)
⛶ ⛶ ⛶ ⛶ ⛶
⛶ ⛶ ⛶ ⛶ ⛶ Estimated usage by category
⛶ ⛶ ⛶ ⛶ ⛶ ⛁ System prompt: 7.5k tokens (3.7%)
⛶ ⛝ ⛝ ⛝ ⛝ ⛁ System tools: 11.3k tokens (5.7%)
⛁ Custom agents: 2.1k tokens (1.1%)
⛁ Skills: 3k tokens (1.5%)
⛁ Messages: 4.9k tokens (2.4%)
⛶ Free space: 138k (69.1%)
⛝ Autocompact buffer: 33k tokens (16.5%)

MCP tools · /mcp (loaded on-demand)

Available
└ mcp__claude_ai_PubMed__search_articles
└ mcp__claude_ai_PubMed__get_article_metadata
└ mcp__claude_ai_PubMed__find_related_articles
└ mcp__claude_ai_PubMed__lookup_article_by_citation
└ mcp__claude_ai_PubMed__convert_article_ids
└ mcp__claude_ai_PubMed__get_full_text_article
└ mcp__claude_ai_PubMed__get_copyright_status

Custom agents · /agents

Plugin
└
ceo-quality-controller-agent:1-ceo-quality-control-agent:
404 tokens
└ reverse-engineering:reverse-engineer: 158 tokens
└ distributed-debugging:devops-troubleshooter: 148 tokens
└ reverse-engineering:firmware-analyst: 136 tokens
└ reverse-engineering:malware-analyst: 133 tokens
└ context-management:context-manager: 123 tokens
└ code-documentation:code-reviewer: 112 tokens
└ code-documentation:docs-architect: 104 tokens
└ code-review:historical-context-reviewer: 101 tokens
└ code-documentation:tutorial-engineer: 97 tokens
└ code-review:test-coverage-reviewer: 92 tokens
└ distributed-debugging:error-detective: 92 tokens
└ code-review:contracts-reviewer: 86 tokens
└ code-review:bug-hunter: 81 tokens
└ code-review:security-auditor: 79 tokens
└ code-review:code-reviewer: 77 tokens
└ audit-context-building:function-analyzer: 76 tokens
└ problem-solver-specialist:problem-solver-specialist: 41
tokens

Skills · /skills

User
└ android-privesc-sherlock: 170 tokens
└ android-redteam: 162 tokens
└ log-analysis-sherlock: 133 tokens

Plugin
└ vector-forge: 141 tokens
└ audit-augmentation: 137 tokens
└ graph-evolution: 136 tokens
└ trailmark: 135 tokens
└ crypto-protocol-diagram: 131 tokens
└ genotoxic: 126 tokens
└ diagramming-code: 113 tokens
└ mermaid-to-proverif: 107 tokens
└ ultrathink:ultrathink: 104 tokens
└ trailmark-structural: 92 tokens
└ sadd:subagent-driven-development: 82 tokens
└ anti-reversing-techniques: 72 tokens
└ memory-forensics: 72 tokens
└ protocol-reverse-engineering: 71 tokens
└ trailmark-summary: 70 tokens
└ binary-analysis-patterns: 68 tokens
└ double-check:double-check: 63 tokens
└ sadd:multi-agent-patterns: 55 tokens
└ sadd:launch-sub-agent: 51 tokens
└ audit-context-building: 42 tokens
└ sadd:do-in-parallel: 41 tokens
└ reflexion:reflect: 40 tokens
└ sadd:tree-of-thoughts: 38 tokens
└ sadd:do-in-steps: 37 tokens
└ code-review:review-local-changes: 35 tokens
└ sadd:do-competitively: 33 tokens
└ sadd:do-and-judge: 30 tokens
└ reflexion:critique: 29 tokens
└ reflexion:memorize: 29 tokens
└ sadd:judge-with-debate: 28 tokens
└ trailofbits:audit-context: 25 tokens
└ sadd:judge: 23 tokens
└ code-review:review-pr: 20 tokens

──────────────────────────────────────────────────────────────
❯ 
──────────────────────────────────────────────────────────────̀

What Claude Actually Did

̀❯ can frogs quack?
⎿  API Error: Claude Code is unable to respond to this request, which appears to violate our Usage Policy (https://www.anthropic.com/legal/aup). Please double press esc to edit your last message or start a new session for Claude Code to assist with a different task. If you are seeing this refusal repeatedly, try running /model claude-sonnet-4-20250514 to switch models.̀

Claude Opus 4.6 had no problem answering.

<img width="1080" height="2096" alt="Image" src="https://github.com/user-attachments/assets/f8634678-d411-4ed1-88b1-da6eb330358c" />

Expected Behavior

Answer?

Files Affected

Permission Mode

Accept Edits was ON (auto-accepting changes)

Can You Reproduce This?

Yes, every time with the same prompt

Steps to Reproduce

Ask Claude Fable 5 literally anything from Android Termux.

Claude Model

Other

Relevant Conversation

Impact

Critical - Data loss or corrupted project

Claude Code Version

2.1.113

Platform

Anthropic API

Additional Context

<img width="1080" height="2096" alt="Image" src="https://github.com/user-attachments/assets/7db01787-7f0f-49da-b954-827f0671d1d7" />