Writing plugins

Agent skills for cutting AI patterns from writing and adding human voice.

5 skills

Travis Gertz's personal writing voice and editorial style engine

3 skills · 1 command

Review and revise writing with a source-audited, versioned catalogue. Three skills: slop-sense (review and rewrite), slop-check (review only), and slop-explain (pattern explanations). An optional lexical scorer and rhythm checker provide raw measurements separate from editorial findings.

4 skills

Technical presentation decks (pptx) and speaker scripts (md): fixed dark design plus paper-style light-sandwich and dark-step styles

3 skills

기술 블로그 글·기술 회고·아키텍처 설명 글을 플렉스(flex.team) 테크블로그 문체로 작성·교정하는 flex-style 스킬 + 기본 적용용 CLAUDE.md 배선 setup

2 skills

Publish a recurring analysis, such as backlog triage or bug triage, as a live report board with a stable URL, and re-sync it in place as the source data changes.

1 skill

Document-structure guide for multi-milestone formalization roadmaps in Lean, Rocq, Isabelle, HOL, and other proof assistants.

1 skill

Write a numbered, resumable checklist for checks run by hand, where every step says what to do, what you should see, and how to tell nothing from broken.

1 skill

Bilingual (EN/ZH) workflow, LaTeX editing, and prose style rubric accumulated from user feedback on scientific paper writing

1 skill

Skills for removing AI writing patterns and making text sound more natural and human

1 skill

Skill for building and maintaining a paper skeleton (load-bearing sentence per paragraph, roles, argument map) of a LaTeX manuscript in an Obsidian vault

1 skill

Academic writing discipline for HCI students: register, framing, novelty claims, proposal writing, and Master's vs PhD calibration.

1 skill

A forge for Claude Code plugins — academic research, fiction worldbuilding, publication workflows, and more

Smart textbook authoring team: 13-state pipeline orchestration covering instructional design, retrieval, writing, review, and delivery with anti-hallucination citation verification. v2.0: ACP envelope alignment across members. v2.1: task-router 15-type routing + trace/progress observability. v2.6: section-level wave state persistence + contract write-back + lead verdict gate. v2.9: exercise-marker canonical form unified ([习题 {章}-{序}]); legacy form now FAILs 4.8b.

10 skills · 6 agents

Sentence-by-sentence readability filter calibrated to your voice rules. OK / ~ / X grading + clean rewrite.

13 skills

Idea or competitor reel to a short-form script: 10 scored hooks, beat map with re-hooks every 6-8s and proof on screen, quality gate, in your voice.

13 skills

Turn a brain-dump into a full day of ready-to-post Threads/X posts. Proven templates, hook bank, anti-AI-detection filter.

13 skills

Writing gates and working skills, with Hebrew and Israeli-reader modules.

10 skills

Bind every factual claim in a document to a verbatim quote in a real source, then gate on it.

3 skills · 4 commands · 1 agent · hooks

Audit éditorial, playtest, mise en marché et orchestration de pipeline pour suppléments JDR : revue de publication, chasse aux clichés, tics récurrents, noms génériques, simulation de playtest multi-passes, évaluation commerciale (listing, pricing, visuels, positionnement DriveThruRPG), conversion entre systèmes (5e 2014 → 2024 / 5e → PF2e / 5e → OSR / 5e → narratif), audit mécanique de stat blocks 5e 2024, et agent orchestrateur ttrpg-publication-director qui pilote le pipeline complet draft → launch.

8 skills · 1 agent

Prose craft through a matched triple (/author-prose-architecture, /author-prose, /audit-prose) over caller-supplied kind layers: interface, documentation, copy

6 skills · 1 agent

Find the tells in your tale. TellTale reads fiction at two levels, the sentences and the story underneath them, and catches what line edits miss: stakes that never rise, a theme the narrator spells out. It ranks what to fix first, for one chapter or a whole manuscript in Markdown, Word, EPUB, or PDF. A craft signal, not an AI detector.

1 skill · 4 commands

Four composable skills for planning, scripting, designing, and delivering outstanding keynote presentations.

4 skills

文章スタイルルールをセッション開始時に自動注入し、文章の書き込みを直前に検証して違反を拒否する執筆ガードレール。ルール1項目 = 1サブエージェントの並列全文検証も提供する

1 skill · 1 agent · hooks

Business Russian for client-facing, legal and documentation texts: the principles, the self-check, the checklist, and a review command.

1 skill · 2 commands

Adapts how Claude writes for you: the level of English, the domain knowledge it assumes, and the amount of professional jargon - never the facts.

1 skill · hooks

เขียนข่าว AI/เทคโนโลยีแบบมืออาชีพ โครงสร้างพีระมิดหัวกลับ รองรับไทย/อังกฤษ — Write professional AI & tech news in Thai or English.

1 skill · 1 command

Two skills for AI-drafted technical communication: deslop-tech-comms (proactive draft-time: audience triage, surface-specific structure, anti-slop checklist for docs, summaries, Slack posts, emails, status updates) and thermo-nuclear-writing-review (reactive review-time: strict prose audit parallel to thermo-nuclear-code-quality-review, with writing-judo moves, length discipline, prose-spaghetti detection, AI-fingerprint and buzzword bans, APPROVE/REVISE/MAJOR REVISION verdict). The pair shares one vocabulary so draft-time and review-time passes don't fight each other.

2 skills

Audits text for AI writing patterns and rewrites it to sound human. 12 rule IDs, separate EN and PT-BR catalogs, regression suite.

1 skill

Model- and harness-agnostic personal-voice corpus builder. Pulls your own writing from any source (iMessage, Slack, Discord, email/mbox, your own articles, or any text export) through one shared cleaning + anti-AI-slop + length-tiering pipeline, then emits a JSONL corpus plus a portable 'voice pack' prompt that makes any LLM write in your real voice. Pure-stdlib core; new sources are a one-file adapter. Includes a research writeup on why corpus exemplars beat prompting and when a local fine-tune wins instead.

1 skill