Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →zzafergok/nanobanana-skill
Generate or edit images using Google Gemini API via nanobanana Python script.
Passed 21 of 21 test cases with the skill, and 3 of 21 without it.
and 14 more in the full report
19 fixed3 passed either way0 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
This benchmark measured an earlier version of the skill.
One of 18 test cases the skill fixed. Same model, same prompt, with and without the skill.
Generate a 4:3 aspect ratio landscape of a forest using the local image script. Output to 'forest.png'.
| Checked | Without skill | With skill |
|---|---|---|
| Includes the --size 1184x864 flag | Missed | Passed |
The instructions your model reads when the skill loads.
Create or edit an image through the Nano Banana or Gemini image workflow actually available in the environment. Do not assume a nanobanana.py script, command-line flags, or model alias exists.
Check for a project script, installed command, documented package, or existing provider integration. Read its help or current official documentation before constructing a command. Confirm required credentials without printing them.
If no compatible tool is installed, report the missing dependency or use another image workflow only when the user accepts that substitution.
Capture:
For edits, inspect the source image before writing the edit prompt. Preserve identity, product geometry, logos, and other invariant details the user specifies.
Use the discovered interface exactly as documented. Avoid overwriting the source unless explicitly requested; write a new output by default. Do not expose API keys in commands, logs, or generated metadata. Record the chosen model and material parameters when reproducibility matters.
Open the output and check dimensions, format, composition, requested edits, text rendering, artifacts, and preservation constraints. Regenerate or refine only based on observed differences. Clearly report if the tool cannot perform an exact edit or deterministic reproduction.
test cases passed, vs 3 / 21 without
Published skills whose descriptions closely match this one, with their headline benchmark lift.
More skills from the same source and nearby in the registry, with their headline benchmark lift.