Skip to content
Back to search

fablize

MarketplaceWorks with Claude CodeGitHub

A Claude Code plugin that makes Opus behave like Fable — completion, evidence, and verification enforced as procedure. Ships only what a Fable-vs-Opus comparison proved transferable.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/fivetaku-fablize/badge.svg)](/listings/fivetaku-fablize)

HTML

<a href="/listings/fivetaku-fablize"><img src="/api/listings/fivetaku-fablize/badge.svg" alt="findsafeskills safety rating" /></a>

Install

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

/plugin marketplace add fivetaku/fablize
/plugin install fablize

The per-task router (a UserPromptSubmit hook) registers automatically.

For always-on operation (the rules resident in context), run once:

bash ${CLAUDE_PLUGIN_ROOT}/setup/setup.sh    # choose local (recommended) or global

Uninstall: `bash ${CLAUDE_PLUGIN_ROOT}/setup/uninstall.sh`

> If fablize helps, a ⭐ on the repo means a lot — setup asks once (a single opt-in question); it never stars without your explicit "yes".

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 16 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. {
  2. "name": "fablize",
  3. "owner": { "name": "fivetaku", "email": "gptaku.ai@gmail.com" },
  4. "metadata": {
  5. "description": "fablize — a harness that makes Opus behave like Fable (completion, evidence, verification as procedure)",
  6. "version": "2.1.1"
  7. },
  8. "plugins": [
  9. {
  10. "name": "fablize",
  11. "source": "./",
  12. "description": "A harness that makes Opus behave like Fable. Ships only the verified-effect procedures distilled from a Fable-5-vs-Opus-4.8 comparison: render-output verification, multi-story evidence gate, investigation protocol, early-stop guard, plus per-task routing. Capability gaps are not faked — see README.",
  13. "category": "workflow"
  14. }
  15. ]
  16. }
README.md · README · 78 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. English | [한국어](README.ko.md)
  2. # fablize — run Opus like Fable
  3. [![GitHub stars](https://img.shields.io/github/stars/fivetaku/fablize?style=flat&color=yellow)](https://github.com/fivetaku/fablize/stargazers)
  4. [![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](https://opensource.org/licenses/MIT)
  5. A Claude Code plugin that makes Opus (or any Claude model) **see a task through to the end — with evidence and verification — as procedure, not as luck.**
  6. ## Why
  7. When Fable 5 shipped, I ran a controlled comparison of Fable 5 and Opus 4.8 (an A/B set of 19 runs plus 26 real working sessions, ~1,500 tool calls). The finding:
  8. - On closed, answer-bearing work (code, logic, builds), the two models were effectively tied.
  9. - The gap appeared only on open-ended work, and its nature was "following an implication one step further."
  10. - That depth is **model capability** — it could not be transferred by instructions or a harness. An injection experiment confirmed it: Opus could not reproduce the defects Fable found on its own.
  11. - But the **procedure** of good work — actually running what you build, seeing it through, investigating systematically — does transfer.
  12. fablize applies only the procedures whose effect was verified. It does not raise the model's ceiling; it makes the model reach its own ceiling.
  13. ## What transfers and what doesn't
  14. | Trait | Transferable? | Reason |
  15. |---|:--:|---|
  16. | Verification grounding (run & observe the artifact) | ✅ shipped | A procedure Opus skipped "not because it can't, but because it didn't." Injecting it raised render-verification behavior measurably. |
  17. | Multi-story completion + evidence gate | ✅ shipped | A procedure — decompose, checkpoint, refuse completion without proof. |
  18. | Systematic investigation (reproduce → hypotheses → causal chain) | ✅ shipped | A procedure — on par with what the strong model already does. |
  19. | Early-stop prevention | ✅ shipped | A deterministic hook — catches "I'll do X" without doing it. |
  20. | Out-of-spec defect discovery | ❌ not possible | Capability. Injection was refuted — the model finds it, or it doesn't. |
  21. | Open-ended creative detail | ❌ not possible | Capability. Shows only where there is no fixed answer. |
  22. | Self-driven propagation depth | ❌ not possible | Capability. Directed propagation transfers; self-started depth does not. |
  23. The non-transferable rows are the model's job (or a human's), not a harness's. When you hit them, fablize tells you to escalate instead of pretending.
  24. ## What's included (verified only)
  25. - **Verification grounding** — render/executable artifacts (HTML, SVG, games, charts) are run and observed before completion.
  26. - **Multi-story verification gate** — `goals.py` decomposes work and refuses a groundless "done."
  27. - **Investigation protocol** — reproduce, compete hypotheses, trace the full causal chain.
  28. - **Early-stop hook** — blocks promising-without-doing.
  29. - **Per-task router** — injects only the matching verified discipline.
  30. Negligible or unverified ideas (style mimicry, broad reasoning injection, a silent-recovery guard, a review-recall scan) are **not shipped.** They stay in personal development until a controlled experiment confirms their effect.
  31. ## Install
  32. ```
  33. /plugin marketplace add fivetaku/fablize
  34. /plugin install fablize
  35. ```
  36. The per-task router (a UserPromptSubmit hook) registers automatically.
  37. For always-on operation (the rules resident in context), run once:
  38. ```
  39. bash ${CLAUDE_PLUGIN_ROOT}/setup/setup.sh # choose local (recommended) or global
  40. ```
  41. Uninstall: `bash ${CLAUDE_PLUGIN_ROOT}/setup/uninstall.sh`
  42. > If fablize helps, a ⭐ on the repo means a lot — setup asks once (a single opt-in question); it never stars without your explicit "yes".
  43. ## How it behaves
  44. - Trigger with `/fablize` or phrases like "see it through", or automatically when installed always-on.
  45. - 2+ stories → decompose + verification gate. Debugging → investigation protocol. Render artifact → verification grounding. Hard task → adaptive thinking plus an `/effort xhigh` suggestion.
  46. - At the capability ceiling, escalate to a stronger model or a human.
  47. ## Honest limits
  48. - It cannot raise model capability. Open-ended creative quality and self-driven discovery are out of reach — that is a model-choice decision, not a harness one.
  49. - The effect numbers come from a small, single-family (Claude) self-measurement (the 19-run A/B set). The direction is solid; the decimals are not asserted.
  50. - The early-stop hook can misfire on a declarative offer ("I'll write the report if you want") — phrasing offers as questions avoids it.
  51. ## License
  52. MIT — see [LICENSE](LICENSE).

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.