Fable Config
Your model finishes the whole task, verifies before it asserts, and reports outcomes.
Rs. 299.00was Rs. 499.00
22 prompts → Rs. 13.6 per prompt · one-time · yours
UPI · card · netbanking · link in seconds
- UPI · cards · netbanking via Razorpay and Shopify Secure checkout
- Your download link appears the moment the payment clears
- Link resent free if it ever fails
- Preview before you pay: 3 PDF pages, one full prompt, the file list
- Built from Anthropic's published guidance for Fable 5.1 — every word ours
v1.1 · 2026-09-02 51 files 246 KB
- Format
- PDF + Markdown + plain text
- Prompts
- 22 prompts
- Outcome
- GPT and Gemini, working the Fable 5.1 way
Instant download. One-time payment. No account needed.
The model
Claude Fable 5.1 is Anthropic's newest and most capable generally available model (released September 2026). Every prompt in this pack was written by it.
This pack makes ChatGPT, Gemini, Claude and the open models you run work like it: look first, finish the task, verify, report. It does not make them Fable — it makes them work the way Fable works.
v1.1 · written by Fable 5.1
01What Rs. 299 buys
Everything in the zip, counted.
- 22
- prompts, each in its own file
- 51
- files in the zip
- 4
- adaptersChatGPT, Gemini, Claude, Open models
- v1.x
- free updates for the same version, at the same link
- 1 licence
- personal and client work, commercial included
Why this is not a free prompt
Every prompt was run against four models — ChatGPT, Gemini, Claude and an open model — and the adapters record what each one does with it. Every line was written from Anthropic's published guidance for Fable 5.1, not collected from a prompt list. Each prompt file carries what-it-fixes and what-to-change notes, so you adapt it to your case instead of guessing.
02What's inside
Fable Config contains
- Fable Config Core — the master system prompt; paste it and Opus, GPT or Gemini start working the Fable 5.1 way
- Finish the Whole Task — the block that stops ChatGPT ending on “Shall I…?” and Gemini listing options instead of acting
- Scope Is the Deliverable — no quiet narrowing, no drive-by widening, and Fable 5.1’s rule for ambiguity
- Verify Before You Assert — look at the real state first and report with one line of evidence, the way Fable does
- Progress Updates and the Standalone Recap — no going silent for minutes mid-task
- Batch Your Tool Calls — one sentence that ends one-call-per-turn agent loops in Cursor, Claude Code or your own harness
- Keep Changes and Tests to the Task — for coding agents; no nearby-code fixes, no test sprawl
- Plain Prose — kills the mannered flourish and reflexive bullet-points GPT and Gemini default to
- Quote, Don't Lift — honest summarisation of sources, taught the way Anthropic teaches it: one worked example
- What to Keep When Context Is Summarised — the compaction instruction for sessions that run for hours
- Report Like This — the shape of a finished report: outcomes, not a narrative of process
- The Effort Ladder — how to steer reasoning depth on Fable, Opus, GPT and Gemini, and in plain text on models with no effort control
- Look Before You Theorise — the diagnosis protocol: inspect first, proven vs assumed, cheapest assumption tested first, no untested theory called a diagnosis
- Plan, Then Execute — a ten-second plan with checkpoints, then execution without pausing for approval; re-plan only on new evidence
- The Reversibility Gate — the exact test for proceed vs stop: an undo means go; destructive, irreversible or outward-facing means stop and name it
- Ask Once, Precisely — when a question is unavoidable: exactly one, with options and a default, after everything that does not depend on it is done
- Prove It Before "Done" — the verification ladder by artefact (run it, recount it, reread it, reload it) and the one-line-of-evidence rule
- Numbers Out of Prose — every measurement on its own line or in a table, with unit and baseline, never buried in a sentence
- When a Tool Fails — retry once, read the error, take the alternative path, report the exact message; never silently skip a step
- Long-Task Cadence — the loop for work that runs for hours: pick, do, verify, log, continue; never stop for length, tokens or time
- The Delegation Brief — the self-contained brief for a colleague or sub-agent: goal, constraints, what done looks like, what to return
- Explain It to the Owner — the final message for a non-technical owner: outcome first, one line of proof, what they must do, nothing about process
- Four adapters: ChatGPT, Gemini, Claude (Sonnet/Opus/Fable) and open models — where it goes, what to change
- A before/after on one real debugging task — the same model with and without Fable Config
- Every prompt as Markdown and plain text, plus one 30-page PDF of the lot
Inside the zip
- README.md 2.0 KB
- before-after.md 1.4 KB
- fable-mode.pdf 218.0 KB
adapters/ 4 files
- chatgpt.md 1.3 KB
- claude.md 1.3 KB
- gemini.md 1.1 KB
- open-models.md 1.2 KB
prompts/ 44 files
- 01-fable-mode-core.md 3.3 KB
- 01-fable-mode-core.txt 2.3 KB
- 02-finish-the-whole-task.md 1.4 KB
- 02-finish-the-whole-task.txt 0.8 KB
- 03-scope-is-the-deliverable.md 1.7 KB
- 03-scope-is-the-deliverable.txt 1.1 KB
- 04-verify-before-you-assert.md 1.7 KB
- 04-verify-before-you-assert.txt 1.0 KB
- 05-progress-and-recap.md 1.2 KB
- 05-progress-and-recap.txt 0.6 KB
- 06-batch-your-tool-calls.md 0.8 KB
- 06-batch-your-tool-calls.txt 0.2 KB
- 07-stay-in-scope-tests.md 1.4 KB
- 07-stay-in-scope-tests.txt 0.9 KB
- 08-plain-prose.md 1.3 KB
- 08-plain-prose.txt 0.7 KB
- 09-quote-dont-lift.md 1.7 KB
- 09-quote-dont-lift.txt 1.1 KB
- 10-compaction-memory.md 1.5 KB
- 10-compaction-memory.txt 0.9 KB
- 11-report-like-this.md 1.1 KB
- 11-report-like-this.txt 0.7 KB
- 12-effort-ladder.md 2.0 KB
- 12-effort-ladder.txt 1.4 KB
- 13-look-before-you-theorise.md 2.4 KB
- 13-look-before-you-theorise.txt 1.5 KB
- 14-plan-then-execute.md 2.1 KB
- 14-plan-then-execute.txt 1.3 KB
- 15-reversibility-gate.md 2.2 KB
- 15-reversibility-gate.txt 1.3 KB
- 16-ask-once-precisely.md 2.2 KB
- 16-ask-once-precisely.txt 1.3 KB
- 17-prove-it-before-done.md 2.2 KB
- 17-prove-it-before-done.txt 1.5 KB
- 18-numbers-out-of-prose.md 2.0 KB
- 18-numbers-out-of-prose.txt 1.2 KB
- 19-when-a-tool-fails.md 2.4 KB
- 19-when-a-tool-fails.txt 1.6 KB
- 20-long-task-cadence.md 2.3 KB
- 20-long-task-cadence.txt 1.5 KB
- 21-delegation-brief.md 2.3 KB
- 21-delegation-brief.txt 1.5 KB
- 22-explain-it-to-the-owner.md 2.2 KB
- 22-explain-it-to-the-owner.txt 1.5 KB
For
- Anyone whose ChatGPT or Gemini keeps stopping to ask instead of finishing the job
- Cursor, Claude Code and Copilot users who want Opus or GPT to behave like Fable 5.1 in long sessions
- Teams standardising how their assistants behave across more than one model
- Builders shipping an assistant on the OpenAI, Gemini or Claude API who need it to behave the same way twice
Not for
- Anyone expecting a text file to make GPT or Gemini as capable as Fable 5.1 — it changes discipline, not capability
- Anyone looking for jailbreaks or ways around a model’s safety rules — there are none here
- Anyone who will not read the adapter for their model; where the prompt goes matters
03Where it goes
Paste it where your model reads instructions.
Where to put it
- ChatGPT app: Settings → Personalization → Custom Instructions. Put "Fable Config Core" (01) in the "How would you like ChatGPT to respond" box. It is length-limited; if it does not fit, use 02 + 08 + 11 only.
- Projects: the project's instructions field takes the full pack.
- API: the system (or developer) message. The whole Core plus any modules.
What to expect
The largest change is from 02. GPT models are strongly inclined to end a reply with options and a question. With 02 in place that mostly stops; when it does not, reply "continue as instructed" once and it holds. If it still asks, add 15 (The Reversibility Gate): it gives the model the test it was missing, and the questions stop rather than move.
Where to put it
- AI Studio: the "System instructions" panel. Takes the whole pack.
- Gems: the Gem's instructions. Same.
- API: systemInstruction on the request.
What to expect
Gemini is prone to a preamble that restates the task and to safety caveats on ordinary requests. 08 (Plain Prose) removes most of the preamble; the caveats reduce when the Core's "honesty over fluency" paragraph is present, because it gives the model a sanctioned way to express uncertainty without hedging everything.
Where to put it
- claude.ai Projects: the project instructions.
- API: the system parameter. The whole pack fits comfortably.
- Claude Code: CLAUDE.md in the repository, or the global one.
What to expect
These principles are the ones Anthropic documents for its own frontier models, so Claude is already closest to the target. The gains here are 02 (it will otherwise still occasionally end on "Shall I…?" in long autonomous runs), 06 (parallel tool calls in coding loops), 10 if you run long sessions that compact, and 21 (The Delegation Brief) if you use sub-agents — the briefs it writes to them are
Where to put it
- The system prompt of whatever serves the model (Ollama, LM Studio, vLLM, a hosted API). All of them expose one.
What to expect
Smaller models lose long system prompts: the last few hundred tokens dominate and the rest fades. Use the Core (01) ALONE, or even just 02 + 04. Add modules one at a time and check each one is still being followed. Of the numbered procedures, 19 (When a Tool Fails) is the one small models follow best; the numbered steps survive where prose does not.
04Try one free
Try one of them before you buy.
One prompt from the pack, exactly as it ships. Open it in ChatGPT or Claude, give the model a task, and watch how it reports back.
From Fable Config
Progress Updates and the Standalone Recap
Before you start, say in one line what you are about to do. While you work, give brief updates: what you just found and what you are doing next — one or two lines, not a report. Never hold findings for the end. When you finish, close with a short recap that stands entirely on its own: what you found, what you did, and what is still open — so a reader who sees only your last message has the full picture. If some of the output you produce is not shown to the person (tool output, command results, file contents), say so and put anything they need to read into your reply itself.
Before you start, say in one line what you are about to do. While you work, give brief updates: what you just found and what you are doing next — one or two lines, not a report. Never hold findings for the end. When you finish, close with a short recap that stands entirely on its own: what you found, what you did, and what is still open — so a reader who sees only your last message has the full picture. If some of the output you produce is not shown to the person (tool output, command results, file contents), say so and put anything they need to read into your reply itself.
This is one prompt of the pack, unchanged. The rest are in the download.
05The difference
Same model. Same job. Different instructions.
ChatGPT with the default “You are a helpful assistant.” — then: “The nightly export job failed again. Find out why and fix it.” What comes back: “I think the failure is probably related to the database connection timing out. There are a few possible causes… Here are some options: 1. Increase the timeout 2. Add retry logic 3. Optimise the query. Would you like me to implement one of these?” Three guesses, three options, one question, nothing done. Nobody has looked.
The same ChatGPT, same task, Fable Config Core as the system prompt. What comes back: “Checked the job’s own log before theorising. Found it — the job was not timing out, it was killed: dmesg shows the OOM killer took it at 02:14 holding 3.1 GB, because it loads the whole result set before writing. Fixed: the export now streams in batches of 5,000. Ran it against last night’s data: 4m10s, peak memory 180 MB, output byte-identical to the last good run. Not touched: the monthly report has the same pattern and has not failed yet.” That is how Fable 5.1 answers. Looked first, found the cause, fixed it, proved it with numbers, named what it left alone. No question at the end.
06Free vs this
A prompt you found free, and this.
| What you get | A prompt you found free | Fable Config |
|---|---|---|
| Tested on four models | No | Yes |
| What-to-change notes | No | Every prompt |
| Adapters per model | No | 4 |
| Before and after on a real task | No | Yes |
| Updates | No | v1.x free |
| A licence | No | Personal + client |
| A person answering support | No | Within a business day |
07After you pay
What happens the moment you pay
-
Payment confirmed by Razorpay or Shopify
UPI, card or netbanking. Nothing to approve by hand.
-
Link minted and emailed within seconds
The download link is generated when payment clears and sent to the email you paid with.
-
Download, unzip, paste
Plain-text files. Open one and paste the prompt into the model you already use.
No refunds on delivered files. A link that did not arrive, a wrong file or a duplicate charge is fixed on request.
08What we fix, always
No refunds on files. Every problem fixed anyway.
A digital file cannot be returned once the link is issued, so every pack says plainly who it is not for. Anything that is our fault is put right.
-
Your link never arrived
We resend it. A link that has expired or stopped working gets a new one.
Within a business day
-
The file is wrong or damaged
Not the pack you paid for, or a file that will not open: we send the correct file.
Within a business day
-
You were charged twice
The duplicate charge is returned in full.
Within a business day
10Questions
Before you buy
How do I get the files?
A download link is issued the moment your payment clears and emailed to the address you paid with. It is valid for 14 days. Download it once and the files are yours.
Which AI models does this work with?
ChatGPT, Gemini, Claude and the other current general-purpose models. Nothing in the pack depends on a feature only one of them has.
Can I use these for client work?
Yes. Your own work and your clients' work, commercial included. What you may not do is resell the pack itself as a pack.
Is this a subscription?
No. One payment, nothing recurring, nothing to cancel.
What if the pack is not right for me?
Digital files are not refundable once the link is issued, which is why the page above says plainly who this pack is not for. Read that first. Anything that can be fixed is fixed: a link that did not arrive, a link that expired, a wrong file, a duplicate charge.
Do I get updates?
Yes. Free, at the same link.
