AgentStack
SKILL verified MIT Self-run

Nano Banana

skill-pigfoot-claude-code-hubs-nano-banana · by pigfoot

|

No reviews yet
0 installs
8 views
0.0% view→install

Install

$ agentstack add skill-pigfoot-claude-code-hubs-nano-banana

✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets Used
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

Are you the author of Nano Banana? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Nano Banana

Unified image generation workflow using a fixed Python script with JSON configuration. Eliminates AI hallucinations by avoiding dynamic code generation.

Two modes:

  • Unified Generation: detailed prompt → write JSON config → run generate_images.py
  • Interactive Prompting: vague prompt or "help me craft" → guide user, then generate

When to Use

  • Image generation ("draw", "create", "generate"), slides, presentations
  • Image editing with AI
  • Prompt help ("help me craft", "improve my prompt")
  • Brand styles ("use trendlife style", "notebooklm style")
  • Reproducible generation ("use seed 42", "regenerate that image")

Mode selection

  • Interactive Prompting: prompt help requested OR prompt is `
  • CRITICAL: use ${CLAUDE_SKILL_DIR} prefix without cd — keeps cwd in user's project so relative

paths in config resolve correctly.

  • cd scripts && uv run --managed-python generate_images.py ...
  • uv run --managed-python ${CLAUDE_SKILL_DIR}/scripts/generate_images.py ...
  1. Track progress — monitor progress/results files (background tasks)
  2. Return paths — report generated image locations

5+ slides → automatic background execution. Details: references/batch-generation.md

Config

Minimal:

{
  "slides": [{"number": 1, "prompt": "...", "style": "trendlife"}],
  "output_dir": "./001-feature-name/"
}
  • output_dir MUST be ./NNN-short-name/ (relative path, 3-digit number, kebab-case)
  • No absolute paths, no model field (use env var), no seed/temperature for gpt-image-2

Full schema, format selection, layout field rules, gpt-image-2 examples: references/config-reference.md

Seed & Temperature (Gemini only)

  • "seed 42" → "seed": 42
  • "regenerate with seed 392664860" → use seed from previous results JSON
  • "temperature 0.5" / "temp 1.5" / "more creative" → "temperature" field
  • Priority: slide.seed > config.seed > auto-generate

Same seed + same prompt + same temperature = identical image. Auto-seeds are recorded in {output_dir}/generation-results.json.

Full parsing patterns + reproducibility detail: references/temperature-seed.md

Style Detection

Detect style: "trendlife", style: "notebooklm", or natural language ("use trendlife style").

  • TrendLife: Trend Red (#D71920), automatic logo overlay — see references/logo-overlay.md and

references/brand-styles.md

  • NotebookLM: clean presentation aesthetic. ⚠️ NEVER use "NotebookLM" name/logo in prompts

(trademark) — see references/slide-deck-styles.md

Priority: inline spec → ask in Interactive Mode → no style (default).

TrendLife layout field

| Value | Use for | Logo handling | |-------|---------|---------------| | "featured" | title, divider, closing | Logo as reference image, AI integrates naturally | | "content" | information slides | Post-generation overlay, bottom-right corner | | omitted | (auto-detect from keywords — less reliable) | — |

Always specify layout explicitly for TrendLife. Full detection rules + manual override: references/logo-overlay.md


Interactive Prompting Mode

Trigger: prompt help requested OR prompt <5 words.

| Step | Action | |------|--------| | 1. Gather | Check for reference images, existing prompts, inline style specs | | 2. Clarify | AskUserQuestion: output type? subject? style? | | 3. Select technique | Pick from 16+ patterns in references/guide.md | | 4. Generate prompt | Apply technique + brand style + aspect ratio | | 5. Present | Show prompt with explanation and variations | | 6. Execute | Run via Unified Generation Workflow |

Brand style integration:

  • TrendLife (style: "trendlife"): see references/brand-styles.md
  • NotebookLM (style: "notebooklm"): use "clean professional presentation aesthetic" — never the

"NotebookLM" name. See references/slide-deck-styles.md

Techniques and examples: references/guide.md


References

| Topic | File | |-------|------| | API compatibility, model × feature matrix | references/api-compatibility.md | | Batch / multi-slide generation | references/batch-generation.md | | TrendLife brand specs | references/brand-styles.md | | Config schema, layout field, format selection | references/config-reference.md | | Example workflows (4 scenarios) | references/examples.md | | Prompting techniques (16+) | references/guide.md | | Logo overlay rules + manual override | references/logo-overlay.md | | NotebookLM / infographic / data-viz styles | references/slide-deck-styles.md | | Seed & temperature (Gemini only) | references/temperature-seed.md | | Errors, debugging, RONE limitations | references/troubleshooting.md | | Temperature & seed testing data | ../EXPERIMENT_RESULTS.md |

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet — be the first.

Versions

  • v0.1.0 Imported from the upstream source.