Essay Humanizer (corpus-informed)#

Rewrites AI-generated argumentative/academic essays toward human baseline style informed by CAWSE (M/D bands) LOCNESS, and contrast with DeepSeek-generated counterparts. Ships with a fine-tuned LoRA adapter (9.3 MB) and inference script.

Skill contract#

ComponentPathNotesInference scriptscripts/inference.pyEntry point — humanize() function or CLILoRA adaptersassets/adapters/adapters.safetensors.json12.3 MB base64 JSON; auto-decoded to binary on first runPattern weightsdata/analysis/weights.jsonCorpus-derived, loaded by inference at runtimeDecoderscripts/decode_adapters.pyReconstructs .safetensors binary from JSON (auto or manual)Installerscripts/install_deps.shOne-time: pip install mlx mlx-lm transformers + decodeBase modelQwen/Qwen3-8B-MLX-4bitDownloaded from HuggingFace on first run (~4.5 GB, cached)

Requirements: Apple Silicon macOS with Python 3.9+.

Quick Start#

bash

bash scripts/install_deps.sh          # one-time: installs deps + decodes adapter
python scripts/inference.py --file draft.txt   # adapter auto-decodes if not already done

Or from Python:

python

from scripts.inference import humanize
print(humanize("Your AI-drafted essay text here..."))

Weighted pattern table (descending priority)#

When humanizing, address higher-weight rows first. Weights are data-driven from corpus analysis (Mann-Whitney); zero-weight rows were not statistically significant.

IDWeightCategoryPatternP06_CLICHE_METAPHORS0.1358vocabularyCliche metaphorsP15_EM_DASH_OVERKILL0.1358punctuationEm dash overkillP21_MARKDOWN_ARTIFACTS0.1358formattingMarkdown artifactsP23_TEXTBOOK_BOLDING0.1358formattingTextbook boldingP12_PRESENT_PARTICIPLE_TAIL0.1133rhetoricalPresent participle tailingP10_RULE_OF_THREES0.0806rhetoricalRule of threesP04_AI_VOCABULARY0.0621vocabularyAI vocabularyP14_COMPULSIVE_SUMMARIES0.0598rhetoricalCompulsive summariesP05_EXCESSIVE_ADVERBS0.0540vocabularyExcessive adverbsP13_OVER_ATTRIBUTION0.0529rhetoricalOver-attributionP11_FALSE_RANGES0.0341rhetoricalFalse rangesP17_TRANSITION_OVERUSE0.0001punctuationOveruse of transition wordsP01_UNDUE_EMPHASIS0.0000contentUndue emphasisP02_SUPERFICIAL_ANALYSIS0.0000contentSuperficial analysisP03_REGRESSION_TO_MEAN0.0000contentRegression to the meanP07_REDUNDANT_MODIFIERS0.0000vocabularyRedundant modifiersP08_FILLER_HEDGING0.0000vocabularyFiller hedgingP09_NEGATIVE_PARALLELISM0.0000rhetoricalNegative parallelismsP16_EN_DASH_AVOIDANCE0.0000punctuationEn dash / hyphen misuse for rangesP18_COLLABORATIVE_REGISTER0.0000registerCollaborative registerP19_LETTER_FORMALITY0.0000registerLetter-style formalityP20_INSTRUCTIONAL_CONDESCENSION0.0000registerInstructional condescensionP22_EXCESSIVE_LISTS0.0000formattingExcessive bulleted/numbered listsP24_EMOJI_SYMBOL0.0000formattingEmoji/symbol injection

Syntactic complexity (MDD / ADD advisory)#

Human Merit / Distinction-range writing in CAWSE often shows variable mean dependency distance (MDD); AI prose may cluster more tightly. When humanizing:

• Reference MDD means from analysis: human ~2.333775514332394, AI ~2.4553791855163483.
• Variance ratio (human/AI) ~1.7153931408079544: prefer natural mix of shorter and longer dependency links, not uniformly smoothed sentences.
• Avoid flattening every sentence to minimal dependency length; that can read as a different kind of machine polish.

Mandatory rule (orchestrator)#

1. Output continuous prose suitable for submission (no chat-signoffs, no "hope this helps").
2. Plain text only for math if any — no raw $$ LaTeX unless user explicitly requests LaTeX.
3. Preserve author stance and citations if present; do not fabricate references.

Hosted HTTP API (optional, for non-Mac or remote use)#

For non-Apple-Silicon machines or multi-user deployments, run the optional FastAPI server on a Mac host and connect via HTTP/OpenAPI:

1. Install: pip install fastapi uvicorn[standard]
2. Run: uvicorn api.main:app --host 0.0.0.0 --port 8765 (set HUMANIZE_API_KEY env var for auth)
3. Point MCP / OpenAPI tools at https://<your-host>/openapi.json
4. Call POST /v1/humanize with JSON {"text":"..."} (+ Authorization: Bearer …)

See references/hosted_api.md for details.

References#

• references/patterns.md — 24 pattern details with detection/fix hints
• references/training.md — full training pipeline
• references/hosted_api.md — HTTP API / MCP tool linking