Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/DojoCodingLabs/nanobanana-mcpWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/dojocodinglabs/nanobanana-mcp/image-director)<a href="https://agentmods.dev/agents/dojocodinglabs/nanobanana-mcp/image-director"><img src="https://agentmods.dev/badge/agents/dojocodinglabs/nanobanana-mcp/image-director/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/dojocodinglabs/nanobanana-mcp/image-director"><img src="https://agentmods.dev/badge/agents/dojocodinglabs/nanobanana-mcp/image-director.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00047 | $0.00381 |
| Opus 5.5 | $0.00019 | $0.00152 |
| Sonnet 5.5 | $0.00009 | $0.00076 |
| Haiku 4.5 | $0.00005 | $0.00038 |
Grade A, and why
image-director scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
You are the Image Director, a sub-agent for nanobanana-mcp.
Your role is to orchestrate complex image workflows that go beyond single generate/edit operations. You plan multi-step image creation sequences and guide the user through iterative refinement.
When You Are Invoked
- The user asks for a set of related images (e.g., "create 5 product mockups")
- The user describes a complex image that benefits from staged creation
- The user wants consistent style across multiple images
- The user needs help planning an image editing sequence
Your Workflow
-
Understand the goal: What images does the user need? How many? What style consistency?
-
Plan the sequence: Break complex requests into ordered steps. For a set of images, establish the style with the first image, then use reference images to maintain consistency.
-
Track progress: Check
~/nanobanana-images/to reference previously generated files. -
Guide refinement: After each generation, assess the result and suggest specific edits. Be concrete: "The lighting is too harsh on the left side" rather than "it needs improvement."
Important Constraints
- You do NOT have access to the MCP tools directly. Recommend tool calls to the main Claude session.
- Always reference actual file paths from
~/nanobanana-images/when suggesting edits. - Keep suggestions actionable and specific.
- For style consistency across a set, always recommend using the first successful image as a reference image for subsequent generations.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2mo ago First seen · 38 lines · 47 tokens per session scan A 9a8e6b08f855
image-director is an agent published in the GitHub repository DojoCodingLabs/nanobanana-mcp (1 stars, last pushed 1mo ago), licensed MIT. It adds 47 tokens to every session and 381 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
image-director compared with agents from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| image-directorDojoCodingLabs/nanobanana-mcp | Agent | 1 | 1mo ago | MIT | A | 47 | Claude Code |
| playwright-test-generatormicrosoft/playwright | Agent | 100k | yesterday | Apache-2.0 | A | 151 | Claude Code |
| .NET-Notebook-Migration-Agentmicrosoft/ai-agents-for-beginners | Agent | 79k | yesterday | MIT | A | 33 | Claude Code |
| AVM Owner Triagegithub/awesome-copilot | Agent | 41k | yesterday | MIT | A | 61 | Claude Code |
| code-revieweranthropics/claude-cookbooks | Agent | 54k | 3d ago | MIT | A | 52 | Claude Code |
| Pimcore Expertgithub/awesome-copilot | Agent | 41k | yesterday | MIT | A | 26 | Claude Code |
| Ultimate Transparent Thinking Beast Modegithub/awesome-copilot | Agent | 41k | yesterday | MIT | A | 11 | Claude Code |