AI Agent Skill for Gemini Image Cleanup

Give Codex CLI, Claude Code, or another Markdown-aware agent a repeatable workflow for cleaning local Gemini and Nano Banana image assets while it updates docs, prepares repo files, or runs a batch job.

Install the skill Use online tool

What you get

A reusable agent instruction file plus a local script for cleaning visible Gemini logo watermarks in JPG, PNG, and WebP. Add it to a supported agent workspace; other Markdown-compatible environments can adapt the setup.

Tested with

Verified with Codex CLI and Claude Code. Other agents that load Markdown skill files can adapt the same instructions, but you may need to wire the install command yourself.

Clear limits

The skill only handles visible Gemini Logo and Nano Banana watermarks. It does not do black-box uploads, and it does not guess or rebuild damaged pixels. For one or two images, the online tool is usually faster.

Install the local cleanup skill

Add it to an agent workspace once, then call the bundled script whenever image cleanup is part of a larger task.

agent-terminal
# Install the skill into your agent workspace
pnpm dlx skills add GargantuaX/gemini-watermark-remover \
--skill gemini-watermark-remover

# Clean a local Gemini image from an automation flow
node skills/gemini-watermark-remover/scripts/run.mjs ./input.png ./clean.png

Workflows where the skill pays off

Repo asset cleanup

Let Codex or another coding agent clean generated images while updating docs, examples, or site content.

Batch publishing prep

Process a folder of Gemini outputs before exporting blog covers, social cards, or tutorial screenshots.

Private local pipelines

Keep source images on your machine and combine cleanup with renaming, format checks, or release scripts.

Gemini watermark remover skills FAQ

What are Gemini Watermark Remover Skills?
They are reusable AI agent instructions and scripts that let an agent remove visible Gemini logo watermarks from local image files.
When should I use the skill instead of the online tool?
Use the skill when cleanup is part of a larger agent task or batch workflow. Use the online tool when you only need to drag in one or a few images by hand.
Is this the right page for manual image cleanup?
Usually no. This page is for AI agent workflows. For manual cleanup, use the online Gemini watermark remover on the homepage.
Does the skill upload images?
No. The intended workflow is local processing. Images stay in the agent workspace unless your own automation sends them elsewhere.
Can it remove SynthID or invisible watermarks?
No. It targets the visible Gemini logo watermark only, not invisible signals or watermarks from other tools.
Is the skill different from the CLI?
Same engine, different shape. The CLI is a one-shot command; the skill is an instruction file an agent reads before deciding when to call the CLI. Pair them when an agent should clean images as part of a multi-step task.
Is there an MCP server?
Not yet. The skill calls a local script, which is enough for repo-bound agent workflows. If you need a server-side tool surface, the CLI's exit codes and --json output stay stable.
Is the skill free to install?
Yes. The install command points to the open-source Gemini Watermark Remover project. You can inspect the code before adding it to your agent workflow.

Plug it into the next agent run

Install once per workspace; the agent does the rest. The online remover stays a click away for one-off images.

Install the skill View on GitHub