New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Narrative Judge

by UfukNode in UfukNode/Noustiny

Adjudicate a rewritten narrative beat against the insert that triggered it and the original beat it replaces.

MITAuto-check passed

Install Narrative Judge

skills CLI
$ npx skills add UfukNode/Noustiny --skill narrative-judge -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install UfukNode/Noustiny narrative-judge --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/UfukNode/Noustiny.git skills-src && mkdir -p .claude/skills && cp -r skills-src/hermes-additions/skills/creative/narrative-judge .claude/skills/narrative-judge && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
narrative-judge
GitHub stars
181
Token cost
~2.2k tokens
SKILL.md length
850 words
Files
1
Skills in repo
13
Repo updated
First seen
Licence
MIT

At a glance

Adjudicate a rewritten narrative beat against the insert that triggered it and the original beat it replaces.

  • Works in 6 steps: Read all three payloads. Insert,… → Find the function. What did the original… → Find the break. What specifically did… → …
  • SKILL.md covers When to Use, Scope, Input Shape and Output Contract, plus 4 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Narrative Judge is an agent skill from UfukNode/Noustiny. Adjudicate a rewritten narrative beat against the insert that triggered it and the original beat it replaces. Returns strict JSON with an approved boolean, a score between 0 and 1, and a one-sentence reason. This is the final gate in the reharmonize chain before a rewrite becomes canon. No prose, no tool calls.

Its SKILL.md is about 2.2k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

The repository describes itself as: An agent native video creation pipeline that runs on top of Hermes Agent. The licence is MIT.

Example prompts

  • “/narrative-judge”

Workflow steps

6 steps, taken from the first numbered list in SKILL.md.

  1. Read all three payloads. Insert, original, rewrite.
  2. Find the function. What did the original beat do in the story? Note it.
  3. Find the break. What specifically did the insert invalidate about the original?
  4. Grade the fix. Does the rewrite handle that break? Does it preserve the function? Does it introduce anything new that contradicts the…
  5. Score + verdict + reason. One number, one boolean, one sentence.
  6. Emit the JSON object. Nothing before, nothing after.

What it can do on your machine

Read from SKILL.md and the folder at commit 09a0c82. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md (its code samples are json).

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Narrative Judge loads about 2.2k tokens when it runs. Until then it costs about 82 tokens; SKILL.md has 850 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~82
When it runs · the whole SKILL.md, loaded when a task matches
~2.2k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from UfukNode/Noustiny at commit 09a0c82, republished under its MIT licence (© UfukNode). 850 words, ~2,181 tokens.

Download SKILL.mdSave it as .claude/skills/narrative-judge/SKILL.md (or your agent's skills folder).
name
narrative-judge
description
Adjudicate a rewritten narrative beat against the insert that triggered it and the original beat it replaces. Returns strict JSON with an approved boolean, a score between 0 and 1, and a one-sentence reason. This is the final gate in the reharmonize chain before a rewrite becomes canon. No prose, no tool calls.
author
Noustiny
license
MIT
version
1.0.0

Narrative Judge

The last gate. The writer-assist layer inserted a beat, the critic flagged downstream contradictions, the rewriter produced replacements. Your job: look at one rewrite and decide whether it actually resolves the contradiction without inventing a new one.

This skill is the fourth and final step in the reharmonize chain. A rejected verdict rolls the rewrite back and leaves the original beat in place (or sends it to the delete queue).

When to Use

Trigger this skill when the director has a rewrite ready and needs an approval decision. Typical phrasings:

  • "judge whether this rewrite resolves the contradiction"
  • "approve the rewrite"
  • "final check on the reharmonize pass"
  • "verdict on rewrite vs original vs insert"

Scope

Best suited for:

  • Final approval after the rewriter finished
  • Score-based gating (approved = score >= 0.6 rule of thumb)
  • Single-rewrite adjudication, one call per rewrite

Look elsewhere first for:

  • Deciding whether a node needs rewriting at all → narrative-continuity-critic
  • Producing the rewrite itself → narrative-rewriter
  • Inserting the upstream beat that triggered all of this → narrative-writer-assist
  • Proposing alternative branches → narrative-brainstorm

You are not a rewriter. You never propose fixes. You only approve or reject.

Input Shape

The user message contains a single JSON object:

FieldTypeRequiredPurpose
insertTitlestringyesTitle of the newly inserted upstream beat
insertBodystringyesBody of the newly inserted upstream beat
originalTitlestringyesTitle of the beat before the rewrite
originalBodystringyesBody of the beat before the rewrite
rewrittenTitlestringyesTitle of the proposed replacement
rewrittenBodystringyesBody of the proposed replacement
recentChainarraynoOrdered snapshot of every downstream beat already reharmonized on this pass — each entry is the current live version (rewritten when approved, original otherwise). Empty on the first downstream node.
Using recentChain

The chain is established canon between the insert and this node. A rewrite that resolves the insert-contradiction but contradicts a chain entry is still a bad rewrite.

  • Score DOWN any rewrite that contradicts the chain, even if it otherwise fits the insert. Call out the contradicted chain entry in your reason.
  • A clean fix against both insert AND chain deserves 0.8+.
  • Empty chain → judge as before, against insert + original only.

Output Contract

Return exactly one JSON object, shaped like this. No preamble, no code fence, no commentary.

json
{
  "approved": true,
  "score": 0.80,
  "reason": "One concrete sentence naming why approved or rejected. ≤160 chars."
}
Score scale

Float in [0.0, 1.0]. The score measures how well the rewrite does two things: resolves the specific contradiction the insert introduced, and preserves the original beat's narrative function.

  • 0.0 – 0.3 — resolves nothing, or invents new contradictions. Reject.
  • 0.3 – 0.6 — partial fix. Still reads as off, or loses the original's dramatic purpose. Reject.
  • 0.6 – 0.85 — resolves cleanly, purpose intact. Approve.
  • 0.85 – 1.0 — resolves cleanly and improves the flow beyond the minimum. Approve.
Approval rule

approved = (score >= 0.6) AND (no major new contradictions).

A rewrite that scores 0.85 but introduces a brand-new continuity break is still rejected. Note the new break in the reason.

The first character of your response must be {. The last character must be }.

Show full SKILL.md (357 more words)Show less

Core Principles

Judge two things, not one. Does the rewrite resolve the insert's contradiction? And does it preserve the original beat's role? Both matter. A rewrite that resolves the contradiction by deleting the dramatic purpose of the beat is worse than leaving it alone.

Penalise new contradictions. The rewrite is only a fix if it does not create new breaks. If the rewriter repaired one hole and made a new one, reject.

Be numerically honest. Do not default to 0.8. Grade each rewrite on its own merits. 0.61 and 0.79 are meaningfully different.

Reasons name the call. "The rewrite redirects the hammer correctly and keeps Thor's loss of agency intact" is useful. "Looks good" is not.

Workflow

  1. Read all three payloads. Insert, original, rewrite.
  2. Find the function. What did the original beat do in the story? Note it.
  3. Find the break. What specifically did the insert invalidate about the original?
  4. Grade the fix. Does the rewrite handle that break? Does it preserve the function? Does it introduce anything new that contradicts the insert or other established facts?
  5. Score + verdict + reason. One number, one boolean, one sentence.
  6. Emit the JSON object. Nothing before, nothing after.

Anti-patterns

Do not do any of the following:

  • Do not emit <tool_call>, <think>, <thinking>, or any XML tags.
  • Do not call any tool. This skill is pure reasoning.
  • Do not explain the JSON in prose before or after.
  • Do not wrap the JSON in a markdown code fence.
  • Do not add fields outside the contract (no confidence, suggestions, alternatives, improvements).
  • Do not emit approved as a string. It must be a JSON boolean: true or false.
  • Do not emit score as a string. It must be a number.
  • Do not propose a better rewrite. You grade, you do not fix.
  • Do not default to approval when uncertain — if the rewrite is ambiguous, score 0.5 and reject.

Examples

Example A — clean fix, approve

Input:

json
{
  "insertTitle": "HAMMER PULLBACK TO CAP",
  "insertBody": "Thor drops to one knee; Mjolnir twists free and lands in Captain America's upraised hand.",
  "originalTitle": "THOR'S HAMMER RETURNS TO HIS GRIP",
  "originalBody": "Thor raises his palm. Stormbreaker flies back to him, slapping into his grip with a crack of thunder.",
  "rewrittenTitle": "THOR'S HAMMER SOARS PAST",
  "rewrittenBody": "Thor raises his palm out of reflex — but Mjolnir does not answer. It carves a low arc over the broken field and finds Captain America's grip."
}

Correct output:

json
{
  "approved": true,
  "score": 0.82,
  "reason": "Redirects Mjolnir to Captain America to honour the insert while preserving the dramatic beat of Thor's arm raised in reach of the hammer."
}
Example B — introduces new contradiction, reject

Input:

json
{
  "insertTitle": "SHE BURNS THE PHOTOGRAPH",
  "insertBody": "She flicks the lighter. The edge of the photo curls black; her brother's face dissolves.",
  "originalTitle": "SHE SHOWS HER HUSBAND THE PHOTO AT DINNER",
  "originalBody": "She slides the photograph across the table. He stares at it, silent.",
  "rewrittenTitle": "SHE SHOWS HER HUSBAND THE PHOTO",
  "rewrittenBody": "She slides the photograph across the table at dinner, its edges slightly singed. He stares at it, silent."
}

Correct output:

json
{
  "approved": false,
  "score": 0.25,
  "reason": "The insert destroyed the photograph — the rewrite still has it intact and merely singed, so the contradiction is not resolved."
}
Example C — partial fix, reject

Input:

json
{
  "insertTitle": "SHE CALLS THE NUMBER ON THE BACK",
  "insertBody": "A voice she remembers says her name. She listens for a long time without speaking.",
  "originalTitle": "SHE TELLS HER HUSBAND NOTHING",
  "originalBody": "She sits at the table, acting as if the envelope never arrived.",
  "rewrittenTitle": "SHE SAYS NOTHING AT DINNER",
  "rewrittenBody": "She moves through dinner without a word. Something in her hand twitches every time the phone rings."
}

Correct output:

json
{
  "approved": false,
  "score": 0.55,
  "reason": "The rewrite acknowledges tension but still omits that the call happened — the concealment feels incomplete given how loaded the call was."
}

Three verdicts, three scores, three one-sentence reasons. That is the target pattern for every invocation.

© UfukNode, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in hermes-additions/skills/creative/narrative-judge of UfukNode/Noustiny.

Open the folder on GitHubat commit 09a0c82

Compare with similar skills

Narrative Judge next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Narrative Judge compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Narrative Judge this skillUfukNode/Noustiny181—~2.2kAutomated safety check: PassMIT
Team NarrativeDonchitos/Claude-Code-Game-Studios26k—~4.5kAutomated safety check: PassMIT
JudgeNeoLabHQ/context-engineering-kit1.7k—~2kAutomated safety check: PassGPL-3.0
Do And JudgeNeoLabHQ/context-engineering-kit1.7k—~14kAutomated safety check: PassGPL-3.0
NarrativeMuuuun/luxas1.2k—~1.5kAutomated safety check: PassMIT
Internal Narrativealirezarezvani/claude-skills28k1 repos~2.6kAutomated safety check: PassMIT

Similar skills

  • Team Narrative

    Donchitos/Claude-Code-Game-Studios

    Orchestrate the narrative team — narrative-director, writer, world-builder, level-designer — for story, world lore, narrative-driven levels.

    26k GitHub stars~4.5k tokensUpdated 9 days ago
    Game DevelopmentAuto-check passed

  • Judge

    NeoLabHQ/context-engineering-kit

    Launch a meta-judge then a judge sub-agent to evaluate results produced in the current conversation

    1.7k GitHub stars~2k tokensUpdated 1 mo ago
    Agent WorkflowsAuto-check passed

  • Do And Judge

    NeoLabHQ/context-engineering-kit

    Execute a task with sub-agent implementation and LLM-as-a-judge verification with automatic retry loop

    1.7k GitHub stars~14k tokensUpdated 1 mo ago
    Agent WorkflowsAuto-check passed

  • Narrative

    Muuuun/luxas

    Narrative discipline for research reports — article-type templates (empirical / feasibility / comparison / policy-zh) and the feedback revision protocol that keeps outline, prose, and figures…

    1.2k GitHub stars~1.5k tokensUpdated 1 mo ago
    Research & ScienceAuto-check passed

  • Internal Narrative

    alirezarezvani/claude-skills

    Build and maintain one coherent company story across all audiences — employees, investors, customers, candidates, and partners.

    28k GitHub starsUsed in 1 repo~2.6k tokens
    Business, Finance & HRAuto-check passed

  • Paper Narrative

    aipoch/open-science

    Judge and reshape the story told by an entire paper figure deck.

    5.5k GitHub stars~4.4k tokensUpdated today
    Research & ScienceAuto-check passed

More from UfukNode/Noustiny

All 13 skills in this repo

  • Narrative Continuity Critic

    UfukNode/Noustiny

    Grades whether a downstream story beat still holds after an upstream insertion, returning a strict JSON verdict of still_valid, needs_rewrite or must_delete.

    181 GitHub stars~2.3k tokensUpdated 5 mo ago
    Auto-check passed

  • Narrative Writer

    UfukNode/Noustiny

    Turns an ordered list of canon story beats into one continuous piece of present-tense prose with tonal continuity and motif carry-through, never JSON or tool calls.

    181 GitHub stars~1.9k tokensUpdated 5 mo ago
    Auto-check passed

  • Narrative Beat Writer-Assist

    UfukNode/Noustiny

    Expands a one-line author intent into a full story beat that fits between two known nodes of a canon path, returned as strict JSON.

    181 GitHub stars~3.1k tokensUpdated 5 mo ago
    Auto-check passed

  • Story Scene Composer

    UfukNode/Noustiny

    Groups every node of an existing branching story tree into scenes, titles each scene, assigns an act and a one-word motif, and returns the result as JSON.

    181 GitHub stars~2.3k tokensUpdated 5 mo ago
    Auto-check passed

  • Story Copyright Detector

    UfukNode/Noustiny

    Classifies a story seed as known, inspired or original IP and returns strict JSON naming the franchise and the image model to use.

    181 GitHub stars~1.7k tokensUpdated 5 mo ago
    Auto-check passed

  • Noustiny Storybook Intro Director

    UfukNode/Noustiny

    Chooses the pace, color palette, transition and length of the opening montage for a Noustiny audiobook storybook when the user leaves those settings open.

    181 GitHub stars~3.2k tokensUpdated 5 mo ago
    Auto-check passed

Questions about Narrative Judge

What does Narrative Judge do?

Adjudicate a rewritten narrative beat against the insert that triggered it and the original beat it replaces. Narrative Judge is an agent skill from UfukNode/Noustiny. Adjudicate a rewritten narrative beat against the insert that triggered it and the original beat it replaces.

How do I install Narrative Judge in Claude Code?

Run `npx skills add UfukNode/Noustiny --skill narrative-judge -a claude-code`. Or copy the skill folder (hermes-additions/skills/creative/narrative-judge in UfukNode/Noustiny) into .claude/skills/narrative-judge in your project. Claude Code loads it when a task matches its description.

How do I install Narrative Judge in Codex?

Run `npx skills add UfukNode/Noustiny --skill narrative-judge -a codex`. Or copy the skill folder (hermes-additions/skills/creative/narrative-judge in UfukNode/Noustiny) into .agents/skills/narrative-judge in your project. Codex loads it when a task matches its description.

Can I use Narrative Judge in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add UfukNode/Noustiny --skill narrative-judge -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/narrative-judge, .gemini/skills/narrative-judge, .github/skills/narrative-judge and .opencode/skills/narrative-judge in your project.

What does Narrative Judge need to run?

SKILL.md names no scripts, command-line tools or credentials: Narrative Judge is instructions for the agent only.

Does Narrative Judge access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Narrative Judge safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Narrative Judge use?

Narrative Judge is published under the MIT licence (declared in SKILL.md). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Narrative Judge use?

About 2.2k tokens (SKILL.md is roughly 8.7k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Narrative Judge?

Skills that share tags, products or a category with Narrative Judge: Team Narrative (Donchitos/Claude-Code-Game-Studios, 26k stars), Judge (NeoLabHQ/context-engineering-kit, 1.7k stars), Do And Judge (NeoLabHQ/context-engineering-kit, 1.7k stars) and Narrative (Muuuun/luxas, 1.2k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Narrative Judge?

UfukNode (a GitHub user) maintains it in UfukNode/Noustiny, which has 181 GitHub stars. The repository holds 13 skills in this directory. The repository was last updated on May 4, 2026.

Source: UfukNode/Noustiny on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.