agentleFS
lorena-bordonaba-paulorena-bordonaba-pau/youtube-agent/.claude/skills/image-refinement/SKILL.md

image-refinement

Skill2 starsChanged 47 days agoMITView on GitHub

Edit an image that already exists instead of regenerating it from scratch. Triggered by "change the background", "remove the logo", "make it darker", "same but without the text". In Spanish, "cámbiale el fondo", "quítale el logo", "hazlo más oscuro", "igual pero sin el texto".

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If the change is about size, crop, format or weight, this is not an edit: python3 tools/export_image.py --image PATH --type thumbnail It is …

    line 19

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Image refinement

WHEN

There is an image that is already mostly right and one part needs changing.

The mistake this skill exists to prevent: passing the nearly-good image as a reference to generate_image. The model then draws another similar image and whatever was working is lost. The image to edit goes in as the source, never as a reference.

BEFORE SPENDING A CREDIT

If the change is about size, crop, format or weight, this is not an edit: python3 tools/export_image.py --image PATH --type thumbnail It is local, deterministic and free. It redraws nothing.

EXECUTION ORDER

  1. Open the current image with the image reading tool and name exactly what is wrong. Without that, the instruction comes out vague and so does the result.
  2. One instruction, imperative, only the change

    python3 tools/refine_image.py --image PATH \
      --instruction "change the background to flat cobalt blue"

    The tool already adds "keep everything else exactly as it is". Describing the whole image in the instruction is what makes it redraw.

    --type defaults to preserve, which keeps the original's shape. Only change it when a different shape was actually asked for.

  3. If there is a face, pass it as a separate reference: --ref face_photo.jpg:likeness. The face does not go in --instruction.
  4. Compare source and result by opening both. An editing model changes things it was not asked to more often than you would expect.
  5. Iterate one change at a time. Two changes in the same instruction collide.

SOURCES

source: generated. An artefact, not a measurement. Name the model when delivering.

OUTPUT

The file path, the model that made it, and what changed compared to the source, after looking at both — not an assumption about what should have changed.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.