value-for-fable
Open-source Claude Code plugin that pushes Sonnet toward near-Opus output on diagnosis and writing tasks at roughly one-third the cost per response.
VFF is one of the more honest prompt-framework repos you can read. It publishes its own failed benchmark — the 2026-06-11 single-run "5/6 wins, 257 points" result did not reproduce under a fair baseline, and the README says so outright. The v2 output style (compression rules removed) scores 87.1 against v1's 76.2 on a neutral rubric and ties bare Opus within noise on diagnostics and length-bound writing; Opus still leads 5-7 points on architecture and complex performance diagnosis. If you already run Claude Code on a Pro plan, VFF is a cheap experiment. If you are on Opus and happy, the author tells you not to bother — believe him.
Verified 7d ago · liveness 52/100 · cite: rightaichoice.com/tools/value-for-fable
- Claude Code users on a Pro plan who want Opus-adjacent output from Sonnet
- Cost-sensitive AI engineers running Sonnet on diagnosis and structured writing
- Indie developers who want an Opus-adjacent quality bar without the Opus bill
- Teams that want raw, reproducible prompt benchmarks instead of vendor claims
- Opus users already satisfied with their setup — the repo's own recommendation is not to switch
- Non-technical users who need a plug-and-play product without Claude Code
- Architecture decisions and complex performance diagnosis, where Opus leads by 5-7 points in the project's own benchmarks
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Value-for-Fable if you are already an Opus user who is happy with your current setup — the project's own README advises against switching — or if your work is mostly architecture decisions and complex performance diagnosis, where Opus still leads by 5-7 points.
VFF itself is free under AGPL-3.0, but it assumes a paid Claude Code / Anthropic subscription, so your real cost is the Sonnet token spend rather than the plugin.
VFF is free to install under AGPL-3.0, so the pricing decision is really about which Anthropic model you route to. Per the repo's own cost table, Sonnet is about 40% cheaper than Opus and 70% cheaper than Fable 5 on a per-token basis, and VFF claims output cost around 0.30x of Opus for v2 on routine tasks. That fits solo developers and small teams already on a Claude Code plan; larger organizations that need contractual SLAs, support, or a permissive license should budget for those separately.
In short
value-for-fable — Open-source Claude Code plugin that pushes Sonnet toward near-Opus output on diagnosis and writing tasks at roughly one-third the cost per response. Best for Claude Code users on a Pro plan who want Opus-adjacent output from Sonnet, Cost-sensitive AI engineers running Sonnet on diagnosis and structured writing, Indie developers who want an Opus-adjacent quality bar without the Opus bill. Free to use.
What's new in value-for-fable
Checked 7 days agoAcross the latest 1 update: 1 changelog entry.
What people actually say about value-for-fable — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
1 mentions across 1 source (GitHub) · researched Aug 30, 2026.
Average across the 1 source that answered — each source counts once, not each post.
- +Near-Opus output quality at roughly one-third the cost per response (3x cost efficiency).
- +Fully transparent — benchmark data, cost models, and configurations are public on GitHub.
- +Fable5-style operational patterns drive consistent, structured reasoning in Sonnet.
- +Reminder hook (reminder.sh) re-injects the skill in >400KB sessions to prevent drift.
- +2-pass review agent: Sonnet drafts, optional Opus reviews — balances cost and depth.
- −Requires advanced Claude Code knowledge and manual configuration—not beginner-friendly.
- −Pure reasoning tasks still favor Opus by 5–7 points, so not a universal replacement.
- −AGPL-3.0 license complicates commercial redistribution and integration.
- −Limited community presence—no active forum, tutorials, or third-party reviews yet.
- −Benchmark claims need independent verification; self-published data may be biased.
- • No monetary cost, but setup time and Claude API usage (you pay for Sonnet/Opus tokens) are real expenses.
- • AGPL license may force open-sourcing your own work if you integrate it commercially.
Viability Score
How well maintained and how widely used is value-for-fable? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →Key Features
- Claude Code plugin with session-mode skill triggered by "VFF" or "패블 모드"
- Passive output style mode that applies to every session
- Recommended v2 output style with compression rules removed (bench re-verified 2026-06-14)
- v1 output style preserved as the original for comparison
- 2-pass review agent: Sonnet drafting with optional Opus reviewer override
- Drift-prevention hook (hooks/reminder.sh) injecting reminders when transcripts exceed 400KB
- Hook registered on UserPromptSubmit via hooks/hooks.json
- 8-section operational structure spanning communication, style, effort, tool discipline, verification, code changes, writing, and token economy
- Four-criterion fixed review standard (missing requirements, factual errors, unexplained clues, length overrun)
- Reproducible benchmark harness with raw data published in bench/
- Published methodology and results in bench/RESULTS.md
- Cost modeling per task type (input-heavy coding vs output-heavy writing) in COST.md
- Markdown-based configuration for skills, agents, and output styles
- Plugin manifest and marketplace catalog in .claude-plugin/
- AGPL-3.0 open-source license
About value-for-fable
Value-for-Fable (VFF) is an open-source Claude Code plugin maintained by itsinseong on GitHub under AGPL-3.0. It targets the gap between Anthropic's Sonnet and its higher-priced models: instead of swapping in a bigger model, it injects an operational structure — clue-first hypotheses, measurement before recommendation, conclusion in the first sentence, verification before declaring done — into Sonnet via Claude Code output styles and a skill. The repo's own framing is that this is elicitation, not distillation: it draws out capabilities Sonnet already has rather than copying Opus's reasoning. It ships as a Claude Code plugin with a session-mode skill (skills/itsvff/SKILL.md, triggered by typing VFF, Value-for-Fable, 패블 모드, 가성비 패블, or "sonnet을 fable처럼"), a passive output style that applies to every session, a delegation-only review agent for a 2-pass Sonnet-draft/Opus-review flow, and a hook (hooks/reminder.sh, wired to UserPromptSubmit in hooks/hooks.json) that re-injects the skill once transcripts exceed 400KB. Two output styles ship: vff.md (v1, preserved as original) and vff-v2.md, the recommended version with the compression rules removed after benchmarks showed compression was a quality debt. A reproducible bench harness with raw data and RESULTS.md is published in bench/, and COST.md covers per-task-type savings and routing. As of the 2026-06-14 re-verification, v2 scored 87.1 on a neutral rubric versus 76.2 for v1, and tied bare Opus within noise on diagnostic and length-constrained writing tasks (87.1 vs 86.2 and 84.8 vs 89.4 across two re-scores), while Opus still led by 5-7 points on deep reasoning. It is for developers already running Claude Code who want to keep Opus for the hard cases and route routine work to Sonnet.
Behind the Verdict
The interesting claim in Value-for-Fable is not that prompting beats model upgrades. It is the specificity of what it changes. VFF reconstructs four observable Fable 5 behaviors and writes them into Sonnet's output style: treat clues in the user's phrasing as evidence that narrows the hypothesis space, propose the cheapest measurement before the fix, lead with the conclusion, and verify before declaring done. The README's before/after on "the API sometimes throws 500s after deploy" shows the difference concretely — bare Sonnet lists three equal candidates and ignores the word "sometimes"; Sonnet+VFF reads "sometimes" as the clue that points at race conditions or resource exhaustion rather than config, then asks for one line of server log before recommending anything. Two things earn this repo credit. First, the benchmark discipline. The original 2026-06-11 single-run result was not reproducible, and instead of burying that, the project says so and rebuilt the harness with a fair baseline, neutral rubric, two independent judges, and raw data in bench/ plus methodology in bench/RESULTS.md. Second, the v3 that added compound-diagnosis decomposition was discarded because it showed no effect under the neutral rubric — a project that kills its own release is worth more of your attention than one that ships every idea. The weaknesses are structural, not fixable by a better prompt. Benchmarks are 1-2 runs per task, two judges, Claude models only, and skewed toward diagnosis and advice tasks; the README warns the two re-scores swing about 5 points, so the direction is solid but the decimals are not. Parity is with Sonnet at Sonnet's ceiling — VFF does not raise that ceiling, and the README is explicit that prompt-level elicitation cannot replicate a stronger model's reasoning. Deep reasoning tasks still favor Opus by 5-7 points. Setup requires Claude Code and comfort with CLI plugin installs and markdown config files, and AGPL-3.0 will be a blocker if you ship VFF-derived code inside a proprietary product. Where it fits: teams already paying Anthropic per token who want a defensible reason to move routine diagnosis, structured explanation, and length-constrained writing off Opus. Where it does not: architecture review, hard performance debugging, or anywhere you need deterministic parity and contractual support. Run your own blind comparison using the bench/ harness before you route production traffic — that is exactly what the repo is built to let you do.
Researching value-for-fable? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas value-for-fable actually fits — and what changes day-one when you adopt it.
Installs VFF, enables the v2 output style passively, and pastes a "sometimes 500s after deploy" bug report into Claude Code instead of escalating straight to Opus.
Outcome: Gets a clue-first diagnosis that reads "sometimes" as the discriminator, is asked for one server log line, and spends Sonnet tokens rather than Opus tokens on the first pass.
Uses COST.md to split task types — input-heavy coding versus output-heavy writing — then runs the bench/ harness against a set of their own prompts to check whether v2 holds up.
Outcome: Has a per-task routing decision backed by their own blind comparison rather than the repo's published numbers, and can justify moving routine work off Opus.
Rolls out the plugin across the team with the v2 output style and the 2-pass review agent, keeping the Opus reviewer override for high-stakes changes only.
Outcome: Consistent structured responses across the team, with Opus spend concentrated on review rather than drafting.
Use Cases
- Run Sonnet with the VFF v2 output style on routine diagnosis to cut per-response cost versus Opus
- Blind-test Opus against VFF-optimized Sonnet on your own prompts using the bench/ harness
- Use COST.md to decide which task types to route to the cheaper Sonnet+VFF pipeline
- Apply the 2-pass review agent so Sonnet drafts and Opus only reviews high-stakes changes
- Keep long Claude Code sessions on-structure with the reminder hook once transcripts pass 400KB
- Standardize structured explanations and length-constrained writing across a small dev team
- Benchmark your own prompts against VFF's four-criterion review standard
- Self-host a cost-aware coding assistant stack with no vendor account beyond your Anthropic subscription
Models Under the Hood
as of 2026-09-23
Limitations
- VFF does not raise Sonnet's ceiling — it elicits behavior Sonnet already supports, and the README says so explicitly.
- Parity with Opus is demonstrated on diagnostic and length-constrained writing samples; the benchmark is 1-2 runs per task, two judges, Claude models only, and the two re-scores swing roughly 5 points, so treat the direction (v2 well above v1, v2 near Opus on routine tasks) as solid and the decimals as approximate.
- Deep reasoning tasks still favor Opus by 5-7 points.
- There is no official support, and AGPL-3.0 may restrict use inside proprietary products.
- Setup assumes familiarity with Claude Code, CLI plugin installs, and markdown config files.
as of 2026-09-30
Verification history
We have re-verified value-for-fable 9 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 9 verification passes.
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published value-for-fable tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Open Source
$0
Ideal for
Developers already running Claude Code who want to test whether the v2 output style holds up on their own workloads before routing production traffic to Sonnet
What this tier adds
Starting tier — everything ships in one AGPL-3.0 repository at no charge, including the session-mode skill, both output styles, the review agent, the reminder hook, the benchmark suite, and COST.md
Where the pricing makes sense
The company stage and team size where value-for-fable's pricing actually pencils out — and where peers do it cheaper.
VFF is free to install under AGPL-3.0, so the pricing decision is really about which Anthropic model you route to. Per the repo's own cost table, Sonnet is about 40% cheaper than Opus and 70% cheaper than Fable 5 on a per-token basis, and VFF claims output cost around 0.30x of Opus for v2 on routine tasks. That fits solo developers and small teams already on a Claude Code plan; larger organizations that need contractual SLAs, support, or a permissive license should budget for those separately.
Setup time & first value
How long it actually takes to get something useful out of value-for-fable — broken out by persona, not the marketing-page minute.
Solo Claude Code user: expect 15-30 minutes to clone the repo, install the plugin from .claude-plugin/, and confirm the skill triggers with "VFF" — then a few sessions to feel the style change. Small team: budget an hour or two to agree on v1 versus v2 and whether the Opus reviewer override is allowed by default. Anyone who has never configured Claude Code skills or hooks should add an afternoon.
Switching to or from value-for-fable
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From bare Sonnet in Claude Code: install the plugin and enable output-styles/vff-v2.md so the structure applies passively to every session.
- →From your own prompt templates: run the bench/ harness against your existing prompts first, then adopt the 8-section structure only for the task types where it wins.
- →From Opus-by-default routing: keep Opus for architecture and complex diagnosis, and move routine diagnosis and length-bound writing to Sonnet+VFF after a blind test.
- ↗To bare Opus: drop the plugin and switch the model — nothing in VFF is a proprietary runtime you have to unwind.
- ↗To your own output style: fork output-styles/vff-v2.md, keep the four-criterion review standard, and strip the sections your team does not use.
- ↗To another prompt framework: export your bench/ results and scoring rubric first, since the harness is the part that is hard to rebuild.
Resources & Guides
Tutorials & Learning
YouTube returned 6 videos for “value-for-fable”, and we withheld 6: 6 did not mention value-for-fable. We are showing none, because we could not prove any of them are about value-for-fable.
Official links
Tools that pair well with value-for-fable
Common stack mates teams adopt alongside value-for-fable, with the specific reason each pairing earns its keep.
Refact.ai
Open-source autonomous AI coding agent that plans, executes, and deploys tasks inside VS Code and JetBrains IDEs.
DevoxxGenieIDEAPlugin
Free, open-source AI code assistant plugin for JetBrains IDEs with spec-driven development and agent mode
Wesight
Open-source desktop workspace that runs Claude Code, Codex, OpenClaw, and other CLI coding agents side by side with per-task model routing.
Featured Head-to-Head Comparisons
Value For Fable vs Bito
Choose Value-for-Fable if you're an indie developer or cost-conscious engineer who wants near-Opus quality from Sonnet without breaking the bank. Choose Bito if you're in a multi-repo enterprise environment where system-wide context and cross-repo dependency analysis are critical for AI coding agents. They serve different needs: one optimizes model prompting for cost/quality, the other provides an architectural context layer.
Value For Fable vs Cognition Ai
Value-for-Fable is a strict cost-optimization play for teams already using Claude Sonnet: it sacrifices turnkey polish for 70% cost savings and Opus-like quality via structured prompting. Cognition AI’s Devin is a full autonomous engineering platform for enterprises, with deep integrations, native cross-platform support, and a $10M productivity guarantee—but at a far higher price point. Choose Value-for-Fable if you have the technical chops and want to maximize Sonnet’s value; choose Cognition AI if you need a self-driving engineer for complex, production-grade tasks.
Value For Fable vs Poolside Ai
Choose Value-for-Fable if you're a cost-conscious developer or small team wanting Opus-like quality at Sonnet prices, and you can self-host under AGPL-3.0. Choose Poolside AI if you're an enterprise in a regulated industry needing custom, secure, on-prem foundation models with multi-agent orchestration and extensive governance—and you have the budget and willingness to engage in a sales process.
Mastra vs Value For Fable
If you need a full-stack agent framework with durable workflows, observability, and multi-agent orchestration, Mastra is the way to go — it's built for production. If you're on a tight budget and want to squeeze Opus-like reasoning from Sonnet with a structured prompting approach, Value-for-Fable gives you that at zero cost, but it's purely a prompting wrapper, not an agent framework. Choose based on whether you need infrastructure (Mastra) or cost optimization (VFF).
Marvin vs Value For Fable
If you're a developer already using Claude Code and want to squeeze Opus-quality output from Sonnet to cut costs, Value-for-Fable is a no-brainer add-on. If you need a flexible Python framework to quickly embed LLM capabilities (classification, extraction, agents) into your own apps, Marvin's decorator approach is more versatile. For non-Python or non-CLI users, neither is a good fit.
Replit Agent vs Value For Fable
If you need Opus-quality reasoning on a Sonnet budget and are comfortable with a self-hosted CLI tool, Value-for-Fable is a game-changer for cost-conscious AI engineers. If you want to build and deploy full-stack apps by typing natural language, with zero setup and a rich IDE collaboration, Replit Agent is the clear choice—especially with its newly reduced cloud pricing and Slack integration. The two tools serve completely different needs: one optimizes prompt quality for Claude, the other automates app creation.
Pieces For Developers vs Value For Fable
If you're an AI engineer trying to squeeze more from Sonnet without upgrading to Opus, Value-for-Fable is a compelling open-source toolkit that delivers statistical parity at lower cost. For developers drowning in context-switches who need a searchable record of code, chats, and meetings, Pieces for Developers offers a polished, privacy-conscious memory layer. They solve orthogonal problems: pick VFF to optimize prompt costs, Pieces to eliminate forgotten context.
Shipixen vs Value For Fable
For AI engineers optimizing Sonnet pipelines, value-for-fable is a no-brainer – free, open-source, and delivers near-Opus quality at Sonnet pricing. Shipixen wins if you need a polished Next.js landing page fast – one-time purchase, 63+ themes, one-click deploy. Pick your poison: reasoning cost savings vs. frontend speed.
Alternatives to value-for-fable
View all
Refact.ai
Open-source autonomous AI coding agent that plans, executes, and deploys tasks inside VS Code and JetBrains IDEs.
DevoxxGenieIDEAPlugin
Free, open-source AI code assistant plugin for JetBrains IDEs with spec-driven development and agent mode
Frequently Asked Questions
Categories
Best-of guides
Used value-for-fable? Help shape our editorial sentiment research.