Fable 5 is included in paid Claude plans only through July 12 — Anthropic extended it five days from the original July 7 cutoff after users pushed back. After the 12th, it moves to usage credits: $10 per million input tokens, $50 per million output tokens. That's the kind of pricing you save for real leverage, not autocomplete.
Most people are spending their remaining window on tasks a cheaper model handles fine. The prompts below are built for the opposite: they turn Fable 5 into an operator that finds the highest-leverage work in your business, delegates it, systematizes it, and hands you back a system instead of a one-off answer.
Copy, paste, fill in your brackets, run.
1. Find Fable-worthy work first
Before you spend the tokens, use Fable to triage. This stops you from burning your window on tasks that were never worth the model in the first place.
You are helping me decide which parts of my work are worth escalating to Claude Fable 5. Do not execute any of the work yet. Your job is to inspect my context, find strong candidates, and turn the best ones into ready-to-run Fable briefs. My role and current goals: [Describe your role, priorities, team, business goals, or operating constraints.] Use this context: [List the sources you can inspect: repositories, project folders, docs, Slack, task lists, email, calendar, analytics, customer research, or anything else relevant.] Constraints: [Deadlines, privacy boundaries, tools you may not use, budget, approval rules.] Build an inventory of: active projects, repeated workflows, stalled decisions, messy backlogs, and work that spans several tools or sources. Score each candidate 1-5 on: multi-source context, delegation fit, judgment required, clear finish line, leverage, and Fable fit. Return the top 10 Fable-worthy use cases ranked, the evidence behind each one, and a ready-to-run Fable Brief for the top three. Stop there — do not start any task until I choose one.
2. Delegate a task overnight
Hand off a job before you log off, wake up to a finished result instead of a status update.
I'm handing you this task to run unsupervised overnight: [describe the task] Done means: [definition of done] Use this context: [documents, access, constraints] Work through the task to completion. If you hit a blocker, do not stop. Use mocks, stubs, or documented assumptions where appropriate. Record each workaround and continue with everything that doesn't require my decision. By morning, leave me: 1. What you completed 2. What you worked around and why 3. What still needs my decision 4. The evidence that the work functions as intended
3. Fix a broken workflow instead of patching the symptom
For the process that keeps failing, running slow, or quietly costing you money every time it runs.
Here is a session log from an agent (or a person) attempting this workflow: [describe workflow] It struggled with: [time, cost, errors, poor outputs, repeated failures] Analyze where the current process, tool, or system breaks down. Find the structural bottleneck instead of patching the latest symptom. Build the smallest reusable improvement: a checklist, template, tool, or system change. Test the upgraded version against a comparable case. Return: 1. The root cause 2. The change you made 3. The before-and-after result 4. What a cheaper process or person can reuse going forward
4. Turn a recurring job into a self-running loop
For anything that happens weekly or more — reporting, triage, follow-ups — that you're still doing by hand.
I want to turn this recurring job into a loop: [Describe the recurring input, desired output, current process, and frequency.] Examples of previous inputs and outputs: [Attach representative examples, including failures and corrections.] Design a loop that: 1. Collects or detects new input 2. Tracks what's already been processed to avoid duplicate work 3. Decides whether the input is actionable 4. Plans and delegates the work 5. Produces or ships the output 6. Verifies the result against explicit standards 7. Routes anything that needs human judgment 8. Retries recoverable failures and records the rest 9. Captures corrections so the next run improves Identify the trigger, schedule, tools, and permissions the loop needs. Build the smallest working version. Return: a written map of the loop, the working version or plan, the human checkpoints, and how learning gets saved between runs.
5. Organize your context so any model can use it
Scattered docs, stale notes, and duplicated instructions make every AI tool worse. This fixes the source, not the symptom.
Audit the context available for this body of work: [Describe the project, role, or recurring workflow.] Current sources: [List folders, documents, meeting notes, style guides, examples, and memory files.] Design a context system that: 1. Gives a concise starting file for any agent 2. Explains what each source contains and when to use it 3. Separates stable rules from temporary project context 4. Identifies conflicts, duplication, stale guidance, and missing information 5. Includes examples of excellent output and known failure modes 6. Moves repeatable procedures into standalone playbooks instead of bloating the main file 7. Defines how new decisions and corrections get saved Return the new context map, what changed, unresolved conflicts, and instructions for keeping it current.
6. Compound a finished run into permanent knowledge
Don't let a good session evaporate. This turns it into something the next run benefits from.
Review this completed session: [Attach the prompt, plan, outputs, evidence, feedback, and final result.] Determine what should be learned from this run. Separate: 1. Durable lessons that apply to future work 2. Project-specific facts that belong in project context 3. Personal or team preferences 4. Process improvements 5. Mistakes that should not become general rules 6. One-off details that should not be saved For every proposed learning, cite the evidence, explain where it should be stored, and show the exact change you propose. Check for conflicts with existing instructions. Return: what you saved, where you saved it, what you deliberately did not save, and how the next run should improve.
7. Build a strategy that survives contact with evidence
For the plan that's currently based on what everyone in the room already agrees with.
Use the attached source pack to analyze [business area, launch, audience, or funnel]. Sources include: [survey data, customer research, analytics, planning documents, meeting notes, internal goals] Our goal is [specific business goal] for [target customer or profile]. Test our assumptions against the evidence. Do not treat internal consensus as fact. Produce: 1. The 10 findings most likely to change how we operate 2. A ranked list of 10 things we should ship, test, or stop doing 3. The evidence behind each recommendation 4. Source conflicts, stale rules, and assumptions that need verification Flag conclusions that depend heavily on one source or internal rule. Stop for me to choose what to act on before executing anything.
8. Turn scattered feedback into one shippable batch
For when complaints and requests are spread across five different channels and nobody's actually acting on them.
Collect feedback about [product, feature, or workflow] from: [Slack channel, support tickets, recordings, production data, customer calls, meeting notes] Group the feedback into themes. Separate: 1. Changes that are clearly actionable 2. Decisions that require my judgment 3. Requests that conflict with our strategy or direction 4. The evidence supporting each theme Keep a record of what's already been processed so feedback isn't duplicated. Create one coherent plan for the actionable changes. After I approve the plan, implement it as one batch. Return what changed, what you skipped, what still needs review, and the evidence that the changes work. Leave the final sign-off to me.
Fable 5 is included in your plan through July 12. After that, every one of these runs on the meter. Pick the highest-leverage one and run it this week.
