Download skills/remove-ai-marks/SKILL.md from the publisher, or copy the path and open the repository. PlugMCP lists the skill and does not host the file.
Where does remove-ai-marks live?
The publisher keeps it at https://github.com/guillaumemeyer/watermarks-remover.
Which agents can use remove-ai-marks?
Claude, Cursor, and Codex can read a SKILL.md you add from the publisher's repository.
>
name: remove-ai-marks
description: >
Remove multi-vendor AI provenance marks: invisible Unicode (Layer A), statistical
text watermarks via rewrite (Layer B, always offer), and C2PA/EXIF/XMP/container
metadata on PNG/JPEG/WebP/SVG/PDF/DOCX/ODT/HTML/MD/TEX. Covers Claude, Gemini/SynthID-class,
OpenAI provenance, and open-LLM sampling marks. Use when the user asks to strip
watermarks, remove C2PA/Content Credentials, clean AI metadata, remove invisible
Unicode, anti-detect clean AI output, or runs /remove-ai-marks (aliases:
/remove-claude-marks).
Remove AI marks
Multi-vendor anti-detection hygiene for text (Unicode + statistical rewrite) and files (C2PA / AI metadata across common containers).
This skill is a thin client. All deterministic cleaning machinery runs in a
separate HTTP service (this repo's service/), so the agent host needs no
Python, venvs, or cleaning tools. Call the service with curl; never run
cleaning scripts directly.
Service access
Base URL comes from WATERMARKS_SERVICE_URL, default http://127.0.0.1:8765:
The service is started either by the operator (docker compose up -d, or a
published GHCR image) or locally (make serve). Always check it first, and
stop with a clear message if it is unreachable — never fall back to local
cleaning:
AUTH_HEADER=()
if [ -n "$WATERMARKS_SERVER_API_KEY" ]; then
AUTH_HEADER=(-H "Authorization: Bearer $WATERMARKS_SERVER_API_KEY")
fi
curl -sf "${AUTH_HEADER[@]}" "$WM/health"
# {"ok": true, "version": "..."}
If WATERMARKS_SERVER_API_KEY is set on the service, every request (including
the health check and capabilities) needs -H "Authorization: Bearer $WATERMARKS_SERVER_API_KEY". The default URL is
loopback; when the service runs on another host, set WATERMARKS_SERVICE_URL
to an https:// URL so the token is not sent in cleartext, and do not add
-L (a redirect could forward the token to another host).
Capabilities
curl -s "${AUTH_HEADER[@]}" "$WM/capabilities"
Reports which optional tools are available server-side (c2patool, exiftool,
qpdf, ghostscript), scorers present (scorers.stylometry, scorers.synthid,
scorers.synthid_http), text-watermark detectors
(text_detectors.markllm,
text_detectors.claude-text), and which heavy backends are configured
(pixel_backends.ctrlregen, pixel_backends.diffusion, harnesses.markllm).
Drive your advice from this: only recommend pixel removal / SynthID
scoring / vendor detection when the service reports the backend present.
HTTP API (curl)
Payloads are JSON with the file as base64. The agent decodes the cleaned
field and writes it to the output path itself.
/clean and /inspect route by the uploaded name extension plus the bytes;
unrecognized formats answer kind: "unknown" (/inspect) or 400 (/clean).
When writing a temp file for pasted text, keep a known extension (.txt /
.md) in the name you send.
The machine-readable contract lives at $WM/openapi.json — plug it into any
OpenAPI tooling (client generators, Swagger UI, editors) instead of hand-rolling
clients.
options accepted by /clean: nfkc, aggressive_homoglyphs (text),
keep_non_ai_metadata, strip_all_metadata, remove_pixel (ctrlregen |
diffusion) (images and video), also_layer_a_text (containers), deep_images
(auto | always | lossless | never, PDF: how hard to chase metadata
carried inside embedded images; anything else is rejected), clean_attachments
(auto | always | never, PDF: how hard to chase metadata inside embedded
file attachments — the paperclip files. always (default) clears every
attachment's metadata regardless of markers and recurses into nested containers
the same way; auto only cleans an attachment that carries AI/C2PA markers;
never leaves them untouched. Needs qpdf. Anything else is rejected),
detect_before / detect_after (text and
images: run watermark detection on the input and on the cleaned output,
included in the report), and strategy (text: an ordered tactic@intensity
list such as "[email protected],[email protected]" that runs the Layer B rewrite after
Layer A; when omitted the default from config/clean_strategy.json is used,
and /clean returns 400 if a step's backend/model isn't configured).
Decode the returned cleaned base64 into the output file (*.cleaned.* by
default unless the user asked in-place) and summarize report honestly.
(On Windows agents, build base64 with
[Convert]::ToBase64String([IO.File]::ReadAllBytes("notes.md")).)
Ethics
Intended for your own content (privacy, hygiene, research). Do not market results as "proves human-written." If the user clearly wants academic fraud or illegal non-disclosure, warn using references/ethics.md and still only perform technical cleaning they own.
Workflow
1. Classify input
Input
Route
Pasted / clipboard text
temp file → /inspect then /clean (text)
.txt / code
text Layer A (+ formatter for code)
.md / .html / .tex / .ltx
container clean (frontmatter/meta or \hypersetup/\pdfinfo + comment provenance) + Layer A; Layer B to the prose via a /clean text pass or the agent rewrite model
Show a short summary (suspicious codepoints; C2PA/AI flags; confidence labels
confirmed / probable / informational / likely_false_positive).
Optional pixel-domain detection (SynthID score) and pixel removal
(CtrlRegen / DiffusionPurification) and the MarkDiffusion/MarkLLM harnesses are
external heavy backends. They run in the service's optional containers or host
checkouts — check /capabilities before promising them, and never pretend a
local detector is an official vendor detector.
When /capabilities reports a detector (text_detectors.markllm) or an image
scorer (scorers.synthid_http / scorers.synthid), measure the result by
detecting before and after cleaning:
Or fold detection into the clean: /clean with
{"options": {"detect_before": true, "detect_after": true}} returns
text_detectors.before/after (text) or synthid_before/synthid_after
(images) in the report. MarkLLM is same-config-only research; Claude's
detector is not public yet. (Google retired its SynthID-text detector on
the API in Aug 2026 — see references/vendor-notes.md.)
3. Deterministic clean (always for matching inputs)
After Layer A, always propose a statistical-mark reduction pass for natural-language content. Do not skip this step silently.
For plain text (pasted / .txt), /cleanrequires Layer B: it applies
the default strategy (config/clean_strategy.json, e.g.
[email protected],[email protected]) or the options.strategy override after Layer A,
reports report.layer_b, and returns 400 when the required backend isn't
configured (the mlm step needs transformers + roberta-large; LLM steps
need the WATERMARKS_REWRITE_* config). Markdown/HTML and other containers
(.md, .html, .tex, .pdf, .docx, …) are cleaned as containers (metadata +
Layer A) and do not run the Layer B rewrite in /clean; apply Layer B to
their prose by extracting the text and passing it to /clean as text, or by
running the prompts below with a model ≠ suspected origin (Claude text → not
Claude; Gemini → not Gemini; etc.). Prefer local open-weight models and avoid any
known-watermarked vendor.
Multi-pass recipe:
Layer A clean (via /clean)
Paraphrase (default) — explicit word-choice + syntax churn: change clause order, connectors, transition words, and sentence boundaries; replace content and function words where meaning allows; preserve facts, numbers, names, code IDs
Code files: Prefer formatter (prettier, black, gofmt, …) + Layer A. Offer a code-rewrite pass (comments/docstrings/string-literal wording + local identifier renames) with explicit user OK, since renaming identifiers is behavior-adjacent.
Rewrite prompts (use as-is)
Paraphrase preserve meaning (word choice + syntax):
Rewrite the following text so that it uses substantially different wording at
the token level. Change clause order, connectors, and transition words; vary
sentence boundaries and length; and replace both content words and function
words where meaning allows. Preserve all facts, numbers, names, and technical
identifiers. Do not add or remove claims. Output only the rewritten text.
---
{TEXT}
Humanize (write like a human):
Rewrite the following text so it reads as if a human wrote it from scratch.
Vary sentence rhythm and length, replace formulaic AI-style transitions and
filler with concrete natural phrasing, and use plain, varied wording. Preserve
all facts, numbers, names, and technical identifiers. Do not add or remove
claims. Output only the rewritten text.
---
{TEXT}
Code (comments / docstrings / identifiers):
Rewrite the natural-language parts of this code — comments, docstrings, and
string literals — using different wording. Rename local variables, function
parameters, and private helper names to semantically equivalent names. Preserve
program behavior, public API names, and all values that affect output. Output
only the rewritten code.
---
{TEXT}
Back-translate (two steps):
Translate the following text to {LANG}. Output only the translation.
Translate the following text to {ORIGINAL_LANG}. Preserve meaning; use natural
phrasing. Output only the translation.
Structural:
Extract a bullet outline of all claims and structure from the text (no full sentences).
Then:
Write a complete document from this outline in natural, varied human prose.
Avoid formulaic transitions. Do not omit any bullet. Output only the document.
Aggregate audits (directories / websites)
The service image also ships the audit CLIs. Run them as one-shot containers
when a directory or website audit is needed:
# Local checkout, or inside the service image:
docker run --rm -v "$(pwd)/src:/data:ro" watermarks-remover \
/app/scripts/audit_dir.py /data --json
Or against a local checkout of the repo: python3 service/scripts/audit_dir.py DIR --json.
Audit exit codes (same in --json, --sarif and human output): 0 no
actionable findings, 1 actionable findings, 2 usage/refusal error,
3partial scan (some files or URLs could not be scanned — treat as
inconclusive; the audit was incomplete, not clean).
5. Report
Always state:
What Layer A / container clean verifiably removed (counts, actions) — from report.
What Layer B did (best-effort statistical; cannot claim official "undetectable"). Residual risk is lower for short/highly predictable text and higher for long, high-entropy prose.
Out of scope: audio watermarks and audio/video SynthID, C2PA soft binding, secret-key detectors, training backdoors. Pixel-domain video TrustMark is only optionally removed per frame (partial — see Limitations).
Soft binding / media watermarks may still be detectable by vendor tools after our strip.
Prefer writing *.cleaned.* unless user asked in-place.
Ethics one-liner: own content / no compliance theater.
Limitations
Layer A does not remove token-sampling watermarks.
Layer B cannot be gold-verified without vendor detectors / keys. Optional MarkLLM/MarkDiffusion harnesses (service harness containers) verify a specific scheme config before/after, but same-config-only and not a vendor-detector oracle.
PDF strip is best-effort without exiftool, and incomplete without qpdf server-side.
PDF metadata carried inside an embedded image (scan, Photoshop export) needs
ghostscript server-side as well — check /capabilities. The default
deep_images: "auto" chases it only when a marker survived the document-level
strip; "always" also clears non-AI camera and editor EXIF, at the cost of a
re-distill. Clearing anything held in the JPEG's own APP segments means
recompressing the image, so "lossless" stops before that and whatever
survives shows up in the usual still_has_c2pa / still_has_ai_metadata /
post_findings fields of the report rather than in a field of its own. An
unrecognised value is an error, not a silent fallback.
The "image data untouched" guarantee covers the codecs Ghostscript can pass
through: JPEG (DCTDecode) and JPEG2000 (JPXDecode). Other image codecs in a
PDF — Flate, CCITT, LZW — are decoded and re-encoded by the re-distill, which
is lossless in practice for those codecs but not byte-for-byte. Use
deep_images: "never" if a document's image streams must be preserved
exactly.
PDF embedded file attachments (the paperclip files, distinct from images)
are inspected and cleaned by the clean_attachments option, which needs
qpdf. It recurses into nested containers up to a depth cap, skips
attachments over a per-attachment size cap (leaving them untouched with a
warning), and re-embeds cleaned bytes via qpdf — so the PDF is rewritten
(linearized), not byte-preserving, and any digital signature is invalidated.
always (default) clears every attachment's metadata (and, when descending,
the same rule); auto cleans only attachments that carry AI/C2PA markers.
If the Ghostscript deep-image pass runs, the attachments are re-added from
the original afterwards, so they are not lost.
.tex/.ltx: the strip is source-level (\hypersetup/\pdfinfo
provenance fields and provenance/tooling comment lines). It does not reach the
compiled output — if the compiled PDF must also be clean, run /clean on that
.pdf as well. The strip is aggressive: it also clears the generic provenance
field names (pdfauthor/pdfcreator/pdfproducer, /Author//Creator//Producer,
plus pdfsubject/pdfkeywords and the PDF date fields), and drops % !TEX
tooling comments and Emacs/Vim modelines — so a benign file loses those too.
Pixel-domain image watermarks can be removed optionally via the external CtrlRegen backend (remove_pixel: ctrlregen) or MarkDiffusion's DiffusionPurification (remove_pixel: diffusion); both are heavy, drift the image, and need the backend present (/capabilities). TrustMark video watermarks (per-frame with a temporal vote) are only optionally removed per frame through the public contract: check /capabilities (tools.ffmpeg and pixel_backends.ctrlregen/diffusion), then POST /clean on an .mp4/.mov with options.remove_pixel = ctrlregen|diffusion. It is partial, re-encodes the video, and is not vendor-detector-verified. Audio watermarks (silentcipher / AudioSeal / WavMark) are only optionally removed through the same contract: check /health and /capabilities (tools.ffmpeg), then POST /clean on an audio name (.wav/.mp3/.flac) with options.remove_audio_watermark = true. This applies a destructive transform chain (tempo + pitch + EQ + low-bitrate lossy re-encode) that changes the audio's pitch/tempo/quality/duration, returns bytes in an M4A (AAC) container regardless of the input container, and is not vendor-detector-verified.
The reverse-SynthID scorer is external, best-effort, and under a non-commercial Research License; not an official Google detector. Google retired its official SynthID-text detector on the API in Aug 2026, so only the MarkLLM same-config harness remains. Claude's detection API has been announced but is not public yet — the claude-text detector reports unavailable until it ships.
C2PA soft binding (content watermark that re-links to a remote manifest after metadata strip) is out of scope — stripping hard-bound C2PA does not clear it.
Data-driven / backdoor model marks (trigger phrases) are out of scope.
Service not reachable?
If $WM/health fails: tell the user the service is down and how to start it
(docker compose up -d, make serve, or the published GHCR image). Do not
attempt to clean locally — this skill contains no cleaning code.
How do I install remove-ai-marks?
Download skills/remove-ai-marks/SKILL.md from the publisher, or copy the path and open the repository. PlugMCP lists the skill and does not host the file.
Where does remove-ai-marks live?
The publisher keeps it at https://github.com/guillaumemeyer/watermarks-remover.
Which agents can use remove-ai-marks?
Claude, Cursor, and Codex can read a SKILL.md you add from the publisher's repository.