v1.5.0: 4 Editorial Modes, Table/Figure Cross-Referencing & Technical Typography Standards

What's Changed in v1.5.0

Major SKILL & Prompt Enhancements

• Sistem 4 Mode Eksekusi Editorial:
  1. Mode Akademik Ketat: Sudut pandang impersonal, pasif ilmiah metodologi terjaga, sitasi terajut gagasan, 100% EYD V.
  2. Mode Esai & Opini: Boleh sudut pandang penulis, ritme variatif, analogi segar, kalimat aktif mengalir.
  3. Mode Bisnis & Eksekutif: Orientasi aksi, buang basa-basi pembuka, konversi paragraf bertele-tele ke poin ringkas terukur.
  4. Mode Koreksi Minimal (Proofread Only): Hanya membetulkan typo, tanda baca, KTSP, dan kata nonbaku tanpa merombak struktur kalimat atau gaya penulis sama sekali.
• Standar Rujukan Tabel & Gambar (Cross-Referencing): Mengganti penunjuk spasial semu (tabel di bawah ini / grafik di atas) menjadi penunjuk bernomor ilmiah (Tabel 4.2 / Gambar 3.1); melarang fabrikasi nomor bila belum ada.
• Standarisasi Tipografi Teknis: Menetapkan aturan tegas kapan memakai monospace (kode, CLI, file, fungsi), italic (istilah asing belum terserap), dan tegak (istilah berpadanan baku).
• Sinkronisasi Total: Terintegrasi penuh pada SKILL.md, PROMPT.md, dan direktori runtime skill Hermes lokal.

Full Changelog: v1.4.0...v1.5.0

v1.4.0: Cross-Domain Style Guide & Multi-Professional Glossary

What's Changed in v1.4.0

Data & References Expansion

• Glosarium Lintas Bidang (45+ istilah): New structured glossary covering 4 professional domains — Software Engineering (tembolok for cache, basis data for database, antarmuka pemrograman aplikasi for API), Law & Regulation (keadaan kahar for force majeure, uji tuntas for due diligence), Journalism (berita sela for breaking news, umpan klik for clickbait), and Corporate Business (pemangku kepentingan for stakeholder, mendayagunakan for leverage).
• Panduan Gaya Lintas Kalangan: New in-depth reference mapping the stylistic register, conventions, and common AI-tells across 5 professional domains: Academic, Tech, Legal, Journalism, and Business.
• PUEBI Official References: Incorporated authoritative subsections on word formation, particles (pun separation rules), compound words (gabungan-kata), and reduplications (bentuk-ulang) directly from Ivan Lanin's PUEBI Daring (CC-BY-SA 4.0).
• Wikipedia-Verified Grammar: Verified 6 core linguistic articles on Indonesian effective sentences, nominalization, verb morphology, and euphemisms to enrich rule explanations.

Full Changelog: v1.3.0...v1.4.0

v1.3.0: Citation Weaving, Epistemic Hedging, EYD V Rules & Annotated Diff Protocol

What's Changed in v1.3.0

New SKILL Capabilities

• Perajutan Sitasi Ilmiah (Citation Weaving): Resolves formulaic in-text citations (Menurut X (2021)... Senada dengan itu, Y (2022)...) into ideational parenthetical citations focused on findings rather than authors.
• Kalibrasi Modalitas Epistemik (Epistemic Hedge Calibration): Maps verbs to certainty levels (empirical, inference, hypothesis) and forbids stacking hedging markers in a single clause.
• Kaidah EYD V & TBBBI: Adds standard Indonesian morphology rules — KTSP nasal assimilation (mempengaruhi $\rightarrow$ memengaruhi, merubah $\rightarrow$ mengubah, mensintesis $\rightarrow$ menyintesis), bound-form compounding (pascabencana, antarmodul, multidisiplin), pun particle rules, and KBBI standard forms (analisis, praktik, risiko, standardisasi, efektivitas).
• Protokol Diff Beranotasi: All edits must now produce a markdown Diff Table (Original / Revised / Rule Cited) so every change is fully auditable.
• Dataset Diperluas: data/kalke-indonesia.json now includes 12 new BAK-* entries for standard morphology and KBBI lexicon (40 total rules).
• Rujukan Resmi Baru: New references/pedoman-kebahasaan-resmi.md summarizing EYD Edisi V (Keputusan Kepala Badan Bahasa No. 0424/I/BS.00.01/2022) and TBBBI Edisi IV.

Full Changelog: v1.2.0...v1.3.0

Humanizer ID v1.2.0: Calque Elimination, Sentence Fusion, and Standalone Prompt

What's Changed in v1.2.0

Major Enhancements

• Calque & Translationese Elimination: Added deep detection for mechanical English loan-translations (e.g. respecting boundaries -> selaras dengan batas, eliminate waste -> meniadakan limbah).
• Light Verb & Nominalization Reduction: Systematically maps bloated light verbs (melakukan pengukuran terhadap -> mengukur, memberikan pengaruh -> memengaruhi) to concise natural verbs.
• Sentence Fusion (Fusi Kalimat Terfragmentasi): Resolves staccato short-sentence rhythm by combining clauses with subordinating conjunctions (karena, sehingga, sedangkan, sebab).
• Safeguard for Scientific Passive Voice & Technical Terms: Prevents aggressive over-correction in academic methodology and preserves established technical loanwords.
• Strict Document Exclusion Zones: Explicitly protects validated research instruments (Likert/survey items) and legal/patent clauses from phrasing alterations.
• Universal PROMPT.md: Standalone prompt block for web users (ChatGPT Custom Instructions, Claude Projects, Gemini Gems).
• Real-world Case Studies: Added realistic side-by-side examples in examples/ directory.

Full Changelog: v1.0.1...v1.2.0

Humanizer ID v1.0.1

Humanizer ID v1.0.1

Rilis perdana panduan editorial untuk menyunting teks akademik bahasa Indonesia dengan mempertahankan makna dan suara penulis.

• 25 pemeriksaan editorial untuk ungkapan klise, struktur kalimat, dan konsistensi istilah. Pola ini bukan bukti kepengarangan AI.
• Prosedur tiga tahap: inventarisasi, sunting minimal, dan bandingkan makna dengan naskah asli.
• Larangan menambahkan data, sitasi, fitur, dan hasil penelitian yang tidak ada dalam sumber.
• Contoh simulasi sebelum–sesudah serta tautan rujukan EYD dan KBBI.
• Lisensi MIT dengan atribusi adaptasi pola dari blader/humanizer (Siqi Chen).

Batasan

Bukan detektor AI, bukan jaminan skor Turnitin, dan belum merupakan panduan yang ditelaah ahli bahasa. Ikuti pedoman tugas atau institusi.

Penggunaan

Baca README dan SKILL.md sebelum memasang skill. Simpan humanizer-id/SKILL.md di direktori skill agen yang aktif.