← Back to all free guides
A free guide by Mariah Brunner ↗ INSTAGRAM ↗ TIKTOK

Steal These
Fable 5 Prompting Hacks

Anthropic published a prompting guide for Fable 5. Here are the three shifts that change how you use it, each with a copy-paste prompt.

Every prompt below is copied word for word from Anthropic’s official Fable 5 prompting guide. I pulled out the three shifts that matter most and when to use each. Fable is more advanced than past models, but you still have to direct it.

Heads Up

Fable 5.1 is now Anthropic’s newest Fable model, and Fable 5 is a legacy model. The guide was written for Fable 5, so test these on 5.1 and keep what helps. On Pro, Fable runs on usage credits. On Max and Ultra you can use up to 50% of your weekly limit on it. (Fable on your plan)

The 3 Hacks

01

Longer turns by default

Anthropic says hard tasks can run for many minutes at higher effort, and calls this one of the biggest shifts teams hit. On an ambiguous task it can over-plan. Add this to keep it moving.

Stop Over-Planning
When you have enough information to act, act. Do not re-derive facts already established in the conversation, re-litigate a decision the user has already made, or narrate options you will not pursue in user-facing messages. If you are weighing a choice, give a recommendation, not an exhaustive survey. This does not apply to thinking blocks.

02

State the boundaries

Fable can occasionally take actions you didn’t ask for, like drafting an email when you were thinking out loud. Tell it what to do and what not to do, so it reports back instead of running off.

Set The Boundaries
When the user is describing a problem, asking a question, or thinking out loud rather than requesting a change, the deliverable is your assessment. Report your findings and stop. Don't apply a fix until they ask for one. Before running a command that changes system state (restarts, deletes, config edits), check that the evidence actually supports that specific action. A signal that pattern-matches to a known failure may have a different cause.

03

Parallel subagents

Fable dispatches parallel subagents more readily than prior models. Lean into it, and tell it when delegating makes sense so it keeps working while they run.

Guide Delegation
Delegate independent subtasks to subagents and keep working while they run. Intervene if a subagent goes off track or is missing relevant context.
Save Them Once

Put these where Fable sees them every time and they keep working. In the Claude app, paste them into a Project’s instructions. In Claude Code, add them to your CLAUDE.md file. Here are all three in one block, plus a bonus from the same guide for long runs.

All-In-One Fable Instructions
Working rules for this project: 1. When you have enough information to act, act. Do not re-derive facts already established in the conversation, re-litigate a decision the user has already made, or narrate options you will not pursue in user-facing messages. If you are weighing a choice, give a recommendation, not an exhaustive survey. This does not apply to thinking blocks. 2. When the user is describing a problem, asking a question, or thinking out loud rather than requesting a change, the deliverable is your assessment. Report your findings and stop. Don't apply a fix until they ask for one. Before running a command that changes system state (restarts, deletes, config edits), check that the evidence actually supports that specific action. A signal that pattern-matches to a known failure may have a different cause. 3. Delegate independent subtasks to subagents and keep working while they run. Intervene if a subagent goes off track or is missing relevant context. 4. (For long autonomous runs) Before reporting progress, audit each claim against a tool result from this session. Only report work you can point to evidence for; if something is not yet verified, say so explicitly. Report outcomes faithfully: if tests fail, say so with the output; if a step was skipped, say that; when something is done and verified, state it plainly without hedging.

Why The Bonus

Anthropic says that in its testing, the progress-audit line nearly eliminated made-up status reports on long runs, even on tasks built to trigger them.

The model got smarter, so your job shifts from typing perfect questions to setting clear rules.

Read Anthropic’s full Fable 5 guide →