Nicheloom

Market intelligence for builders — see what's gaining traction before it's crowded.

Watermarks Remover: Clean LLM watermarks from text and files

Details

External ID
49481577
Source
HN
Company
—
Product
Watermarks Remover: Clean LLM watermarks from text and files
Website domain
github.com
Launched
Aug. 28, 2026
Cohort
—
Upvotes
5
Upvotes percentile
0.12634408602150538
Tags
—
Fetched at
Sept. 10, 2026, 5:31 a.m.
Updated at
Sept. 10, 2026, 5:31 a.m.

Description

Open-source agent skill + stdlib HTTP service that strips invisible Unicode, C2PA/EXIF/XMP, and (best-effort) statistical text marks from files you own. Claude / Gemini-SynthID / OpenAI / Kirchenbauer / keyed-Gumbel class.

Enrichment

Theme
image editing and background removal tools
Vertical
Horizontal
Function
Content generation
Audience
Developer
AI stance
AI feature
Project type
Hobby / open-source project
Normalized one-liner
remove llm watermarks from text and files
Manually corrected
False

Could you build this?

Partial Stripping invisible Unicode and EXIF/C2PA metadata is trivial, but detecting and removing statistical LLM watermarks (Kirchenbauer, Gumbel) requires specialized statistical algorithms.

What it would actually take: The service needs a pipeline stripping document metadata (using ExifTool or standard PDF/image libraries) and zero-width characters, combined with an LLM token-level re-paraphrasing or perturbed sampling module calibrated to disrupt Kirchenbauer green-red list distributions without altering semantic meaning. Implementing robust statistical watermark removal requires NLP researchers familiar with LLM watermarking literature.

Discussion

1 comment analyzed.

Competitors

Other products that read as similar to this one — 88 launches clear the similarity bar, closest 8 shown.

Attention rank: #85 of 89 (itself plus its competitors, highest first — normalized so YC and Product Hunt are compared fairly).

Launched 277 days after the earliest competitor.

Other launches for this product

Same idea, different domain

Nobody's really built a content generation tool for Government yet.