Chen Nano Banana Pro

ReviewAudited by ClawScan on May 10, 2026.

Overview

The image tool’s code is straightforward, but its launch paths and embedded metadata do not match the listed package, so it could fail or run a different local script while using your Google API key.

Before installing, verify that the skill actually runs the included scripts/generate_image.py from its installed directory, not the hard-coded /Users/apple or ~/.codex nano-banana-pro paths. If you proceed, use a limited Gemini API key, monitor quota/billing, and avoid sending confidential images or prompts.

Findings (4)

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.

Concern

ASI04: Agentic Supply Chain Vulnerabilities

What this means

The skill may fail on another machine or, worse, run a different local script at that path if one exists, potentially handling the user's API key, prompt, or images outside the reviewed package.

Why it was flagged

The command uses a hard-coded absolute path for a different skill slug than the supplied registry entry and does not reference the packaged script by a portable relative path.

Skill content

command: uv run /Users/apple/.openclaw/workspace/skills/nano-banana-pro/scripts/generate_image.py

Recommendation

Install only after verifying the command points to this skill's packaged scripts/generate_image.py, and prefer a corrected package with matching registry metadata, _meta.json, and portable paths.

Note

ASI04: Agentic Supply Chain Vulnerabilities

What this means

When run, uv may install newer matching dependency versions, so behavior can change over time based on package updates.

Why it was flagged

The uv script declares runtime dependencies with lower-bound version ranges rather than pinned versions or a lockfile.

Skill content

dependencies = [
#     "google-genai>=1.0.0",
#     "pillow>=10.0.0",
# ]

Recommendation

Prefer pinned dependency versions or a lockfile, especially before using the skill in sensitive or production workflows.

Note

ASI03: Identity and Privilege Abuse

What this means

The tool can act against the user's Google/Gemini API account and may incur quota or billing usage.

Why it was flagged

The skill requires a Google Gemini API key even though the registry metadata declares no primary credential or required env vars.

Skill content

The script checks for API key in this order:
1. `--api-key` argument (use if user provided key in chat)
2. `GEMINI_API_KEY` environment variable

Recommendation

Use a limited Gemini API key via environment variable or a secrets mechanism where possible, and avoid pasting long-lived keys into chat.

Note

ASI07: Insecure Inter-Agent Communication

What this means

Any prompt text and selected input image are shared with Google for processing.

Why it was flagged

For edits, the script sends the selected local image together with the prompt to Google's Gemini image API.

Skill content

contents = [input_image, args.prompt]
...
response = client.models.generate_content(

Recommendation

Do not use private, regulated, or confidential images or prompts unless you are comfortable with Google's API handling and retention terms.