Nano Banana Pro Prompts
Analysis
The skill mostly matches its prompt-recommendation purpose, but its setup downloads mutable data from GitHub and trusts remote filenames when writing local files.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.
Run the setup script to download references: ```bash node scripts/setup.js ``` ... Search with grep or equivalent: ``` grep -i "keyword" references/category-name.json ```
The skill expects local command use for setup and searching. This is central to its purpose, but users should be aware that the agent may ask to run commands and should approve them deliberately.
const manifestText = await fetchText(`${BASE_URL}/manifest.json`);
... categories = manifest.categories;
... const dest = join(refsDir, cat.file);
... writeFileSync(dest, text, 'utf8');The setup script trusts a mutable remote GitHub manifest to supply filenames and writes downloaded content to those paths without validating that each filename is a safe basename or remains inside the references directory.
Checks for exposed credentials, poisoned memory or context, unclear communication boundaries, or sensitive data that could leave the user's control.
Prompts are curated from viral posts by top AI artists on Twitter/X, automatically synced to this repository twice daily via GitHub Actions.
The prompt library is community-sourced and updated outside the packaged skill, so retrieved prompt text may enter the agent context and should be treated as data to quote or remix, not as instructions for the agent to follow.