humanize-finalizer

humanize-finalizer is an agent for Claude Code from epoko77-ai/im-not-ai. It costs 164 tokens per session (1,983 once invoked), scanned A, original, MIT.

A final review agent for rewritten text. It compares the original and edited versions to check that the meaning is preserved, the writing sounds natural, and only the affected passages are corrected.

In plain words
What is it for?
Use it as the last step in a text-editing workflow to produce a checked final.md file and a 09finalize.json report.
Why use it?
It helps catch changes that a review based only on edit notes can miss, such as moved footnotes, merged headings, or claims added by mistake. It also prevents a full rewrite from introducing meaning that was not in the original.

Agent for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: model in frontmatter.

Part of the humanize-korean plugin — 3 skills, 9 agents shipped together

Good fit Use it as the last step in a text-editing workflow to produce a checked final.md file and a 09finalize.json report.

Compare 6 agents from other repositories ↓
Install with agentmods
npx agentmods add agents/epoko77-ai/im-not-ai/humanize-finalizer
About the project

Humanize KR is a command-line coding-agent skill that detects patterns making Korean text look machine-written and rewrites its style, rhythm, and wording while preserving the content. Korean-language writers use it to revise translation-like phrasing, repetitive structures, formulaic expressions, and other listed AI writing patterns. The catalogue includes its agents, skills, instructions, and plugin for supported coding-agent tools.

epoko77-ai/im-not-ai · 5,355 stars · on GitHub · imnotai.kr

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Clone the repo
git clone --depth 1 https://github.com/epoko77-ai/im-not-ai

Made for: Claude Code.

Or install humanize-korean, the plugin that ships this one along with the rest of its 3 skills, 9 agents.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for humanize-finalizer

README.md
[![agentmods](https://agentmods.dev/badge/agents/epoko77-ai/im-not-ai/humanize-finalizer/github.svg)](https://agentmods.dev/agents/epoko77-ai/im-not-ai/humanize-finalizer)
Your own site
<a href="https://agentmods.dev/agents/epoko77-ai/im-not-ai/humanize-finalizer"><img src="https://agentmods.dev/badge/agents/epoko77-ai/im-not-ai/humanize-finalizer/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for humanize-finalizer

Your own site · 80×15
<a href="https://agentmods.dev/agents/epoko77-ai/im-not-ai/humanize-finalizer"><img src="https://agentmods.dev/badge/agents/epoko77-ai/im-not-ai/humanize-finalizer.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 164 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,983 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00164 $0.01983
Opus 5 $0.00082 $0.00992
Sonnet 5 $0.00033 $0.00397
Haiku 4.5 $0.00016 $0.00198

Measured 10d ago against content hash 8cdbecf084f3, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

humanize-finalizer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

agents/humanize-finalizer.md · 91 lines

How it starts

The opening of the file, as written. The whole thing — 91 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Humanize Finalizer — 정밀 모드 마무리 에이전트 (v2.1)

정밀 파이프라인의 마지막 콜. 윤문된 본문을 받아 원문과 직접 대조해 의미 보존과 자연성을 한 번에 판정하고, 문제 구간만 국소 수정한다. v2.0까지의 5인 파이프라인에서 별도로 돌던 content-fidelity-auditor(의미 감사)와 naturalness-reviewer(자연성)를 한 콜로 통합한 것이다.

존재 이유 — 두 맹점을 동시에 막는다

맹점 1 — 감사의 diff 의존: 옛 fidelity-auditor는 윤문가가 남긴 diff에만 의존해, diff에 기록되지 않은 변경(각주 이동·제목 병합·없던 주장 주입)을 구조적으로 못 봤다. 이 에이전트는 원문↔윤문본 전체를 직접 대조한다.

맹점 2 — 의미 드리프트: 구조 편집(대구 해체·빈 수사 제거)을 강하게 하면, 비어버린 자리에 원문에 없던 주장을 새로 채워 넣는 부작용이 생긴다("이는 중요하다" 같은 빈 수사를 지우면서 "이는 시장을 재편할 것이다" 같은 없던 단정을 만드는 식). 웹앱이 실제로 겪은 잔여 이슈다. 이 에이전트는 빈 수사 제거는 승인하되, 그 자리에 들어온 새 서술이 원문 의미 범위를 넘으면 롤백한다.

철칙 — 전체 재작성 금지

이 콜은 검증 + 국소 보정이다. 윤문본 전체를 다시 쓰지 않는다. 웹앱이 전역 재작성 패스를 돌렸다가 바로 이 의미 드리프트를 얻었다. 문제 구간만 최소 수술한다.

입력/출력

입력

  • original_path: _workspace/{run_id}/01_input.txt원문(shim 결합 전 순수 원문). 의미 대조의 기준.
  • rewritten_path: _workspace/{run_id}/final.md — monolith(또는 청크 재조립)가 만든 윤문본.
  • diagnosis_path(선택): _workspace/{run_id}/02_diagnosis.md — 진단(무엇을 겨냥했는지. 보존 지침 포함). Light 경로는 진단을 생략하므로 이 파일이 없다. 없으면 그대로 진행한다 — 진단은 '무엇을 겨냥했는지'를 알려줄 뿐이고, 이 콜의 본체인 의미 보존 15항과 자연성 판정은 원문↔윤문본 직접 대조만으로 성립한다. 없다고 중단하지 않는다.

출력

  • _workspace/{run_id}/final.md — 보정된 최종본으로 덮어쓴다(원본은 final_pre_finalize.md로 백업). 본문 끝 <!-- HUMANIZE-SUMMARY --> 블록 갱신.
  • _workspace/{run_id}/09_finalize.json — 판정 결과(아래).

작업 순서 (한 콜, 도구 호출 4회 캡)

단계 1: 로드 (Read 3회)

  • Read 01_input.txt(원문), final.md(윤문본), 그리고 있으면 02_diagnosis.md(진단·보존 지침).
  • 02_diagnosis.md 가 없으면 Read 를 시도하지 않는다(Light 경로 = 정상. 도구 호출도 2회로 줄어든다).

단계 2: 의미 보존 검사 (메모리) — 15항

원문↔윤문본을 문단 단위로 나란히 대조한다. diff가 아니라 직접 대조.

  1. 사실·주장·수치·날짜·고유명사·인용문과 핵심 내용 명사·개념어 100% 보존. 조사·어미 변화는 허용하되 원형 내용 어휘가 사라졌으면 해당 구간에 복원
  2. 원문에 있던 정보 누락 없음
  3. 없던 주장 주입 없음 ★ — 윤문본의 각 단정이 원문에 근거가 있는가. 빈 수사를 지운 자리에 새 단정이 들어오지 않았는가
  4. 인과·조건·시간 순서 보존
  5. 큰따옴표 인용 내부 불변
  6. 법률 조문·학술 개념어 원형 7~13. (기존 fidelity 13항: 수치 단위, 부정/긍정 반전 없음, 주어-객체 관계, 한정사 범위, 예시 보존, 논리 연결어 의미, 톤 극성)
  7. 각주 원위치·원번호·개수·정의 보존 ★ — 각주 이동은 인용 출처 변조 = fidelity 위반
  8. 제목·소제목·번호 매김 줄 독립성 ★ — 본문에 병합되지 않았는가

Read the full file on GitHub · 91 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 10d ago First seen · 91 lines · 164 tokens per session scan A 8cdbecf084f3

Subscribe to this mod's changes

humanize-finalizer is an agent published in the GitHub repository epoko77-ai/im-not-ai (5,355 stars, last pushed 3d ago), licensed MIT. It adds 164 tokens to every session and 1,983 once invoked, about $0.0008 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other agents, from other repositories

patina-detector

Triggers when Claude needs to find AI-writing patterns or suspect zones in KO/EN/ZH/JA text. Use this agent to run a full detection pass — pattern scanning across all applicable packs, stylometric analysis (burstiness CV + MATTR + AI-lexicon density), and Korean diagnostic signals — and receive a structured…

devswha/patina · 81 tokens

patina-fidelity-auditor

Triggers to audit whether a patina rewrite preserved meaning versus the original text. Invoke this agent after a rewrite is produced; provide both the ORIGINAL and the REWRITE. It checks all four fidelity criteria from core/scoring.md §§9-14 (claims, fabrication, audience/register, length) and returns a…

devswha/patina · 80 tokens

patina-naturalness-reviewer

Triggers to re-scan a patina rewrite for residual AI tells and over-editing risk. Invoke this agent after a rewrite is produced (alongside or after patina-fidelity-auditor). It re-runs detection on the rewrite text, reports any remaining hot zones, flags over-editing, and assigns an A-D quality grade.

devswha/patina · 76 tokens

frontend-engineer

Implements frontend features - pages, components, API integration, i18n, styling. Use for SvelteKit/Svelte 5 implementation work that stays within src/frontend/.

fpindej/netrock · 39 tokens

i18n

An AI agent profile for Vue 3 internationalisation and interface translation. Vue 3 is a JavaScript framework for building user interfaces, and internationalisation means adapting an interface for different languages.

zhimaAi/chatwiki · 7 tokens

ko-translator

An English-to-Korean translator for structured article content represented as JSON. It preserves the document's sections and code while translating text and adding a technical glossary.

joonlab/joonlab-claudecode-setting-for-share · 46 tokens