✦ You asked for this
Free pack · $0Paste one prompt into Claude Code. It finds where the expensive Claude Fable 5 model is worth it in your codebase, and where Sonnet is enough.
Take me to the promptfree · copy and paste✦ Sonnet or Fable
Use Sonnet to decide when to use Fable. If failure costs minutes, use Sonnet. If failure costs days, consider Fable.
Fast, smart, affordable, and usually enough.
Long, risky, expensive to redo, or painful to reason about.
Pick one row from each line for your task and add up the points. The answers go from 1 point (low) to 6 points (high).
| Question | 1 point | 2 points | 4 points | 6 points |
|---|---|---|---|---|
| Cost if it gets it wrong | Annoying, easy to fix | A few hours lost | A day or two lost | Production, customers, or serious money |
| Context it must understand | One file | One module | Several packages or services | Whole system architecture |
| Work length | Under 20 minutes | One focused session | Several stages | Hours or days of agent work |
| Ambiguity | Very clear | Some product judgment | Architecture decisions | Messy and high-context |
| Verification difficulty | Simple test or visual check | Several tests | Regression thinking required | Failures hide in edge cases |
| Your total | Verdict | What to do |
|---|---|---|
| 10 or less | Use Sonnet. This is a pickle jar. | Fable would probably solve it, but that is not the same as being worth it. Start cheaper and escalate only if Sonnet stalls. |
| 11 to 20 | Start with Sonnet. Escalate with evidence. | Run the audit prompt on Sonnet first. If it finds architecture risk, unclear verification, or too much cross-repo context, then bring in Fable. |
| 21 to 30 | Fable is probably justified. | This has enough risk, context, ambiguity, or duration to justify the expensive specialist. Give it a tight task spec and a verification plan. |
✦ Use it now
Run this from your repository root. The output should rank Fable opportunities, pickle jars, Managed Agent candidates, and the first safe prototype.
It gives you a verdict, a ranked list, a pickle jar list, an agent readiness check, and one task spec to try today.
You are auditing this codebase to decide where Claude Fable 5 is actually worth using instead of Sonnet. Important frame: - Fable is expensive. Do not recommend it for routine work. - Sonnet should be the default for contained edits, short summaries, simple debugging, and normal feature work. - Fable is only justified when the task is long-horizon, high-context, ambiguous, risky, expensive to redo, or benefits from multi-stage agent execution. - Use the surgeon/pickle-jar rule: do not hire a surgeon to open a pickle jar. Audit the repository and return: 1. Executive verdict - Should this repo use Fable at all right now? - If yes, for which 3-5 workflows? - If no, what should stay on Sonnet? 2. Ranked Fable opportunities For each opportunity, include: - Workflow name - Why Sonnet may struggle - Why Fable may be worth the cost - Risk if the model gets it wrong - Required files, services, data, and tools - Verification plan - Estimated first experiment under 90 minutes - Decision: Sonnet, Opus, Fable, or Managed Agent 3. Pickle jar list List tasks that sound impressive but should not use Fable. 4. Agent readiness Identify any workflows that need Claude Managed Agents or another agent harness because they require memory, tools, permissions, sandboxes, tracing, or long-running execution. 5. First prototype Give me one copy-paste task spec I can run today in Claude Code to test the highest-value Fable use case safely. Be blunt. Save money where possible. Recommend Fable only when the value is obvious.
✦ Get unstuck
Open Claude Code in your project folder, the repository root. Paste the prompt and send.
The prompt reads a code project, so it needs one. The points guide on this page scores any single task and needs no code.
Sonnet. The idea is to use Sonnet to decide when Fable is worth it, so you do not pay the big-model price just to ask the question.
Start with Sonnet. If it finds architecture risk, unclear verification, or too much context across repos, then bring in Fable.
Five parts: an executive verdict, ranked Fable opportunities, a pickle jar list of tasks that should stay on Sonnet, an agent readiness check, and one copy-paste task spec to test the best Fable use case safely.
The audit flags work that needs Claude Managed Agents or another agent harness when it needs memory, tools, permissions, sandboxes, tracing, or long-running execution.
Email joe@mastermindshq.business. Say which step you were on and paste what you saw.
✦ What next · paid · optional

You came for one piece of my Claude Code setup. If you have your own website, the next useful job for that setup is getting the site found.
Get your website listed on Google and Bing so people (and ChatGPT) can find you by your name and your service. ChatGPT finds websites mostly through Bing, so we tell Bing about you too. Your own Claude Code does the work on your computer, one prompt at a time.
Not for you? Close this page. Everything above stays free.
Created by Joe Che.
ChatGPT is a trademark of OpenAI. Claude is a trademark of Anthropic. This page is not affiliated with either.