MCP server exposing the Veo 3 fashion film director engine as tools.
Three tools with complete input schemas and descriptions, but quality varies. list_visual_models is a simple read-only discovery tool with adequate description. compose_veo3_prompt and generate_veo3_prompt have identical, comprehensive parameter schemas with enum constraints for visual model, dialogue mode, music mode, output type, aspect ratio, and duration. However, descriptions lack actionable context about when to use each tool (compose vs generate distinction is unclear), and output schemas are not documented. Parameter descriptions are present but generic ('Visual model. One of: ...'). No error handling guidance visible. Tool names follow verb_noun pattern correctly.
Assemble the Creative Director system + user prompt for a rough idea, WITHOUT calling any external model. Returns the composed instructions for the host model to execute. Use this to have the assistant itself write the cinematic Veo 3 brief.
Generate a finished Veo 3 prompt by calling Gemini directly. Requires CREATIVE_API_KEY in the environment. If no key is set, use compose_veo3_prompt instead.
List the available Veo 3 visual models and generation controls (visual models, dialogue/music/output options, aspect ratios, durations).
Output schemas not documented. LLMs cannot plan downstream operations or extract structured data from compose_veo3_prompt and generate_veo3_prompt responses.
Weak tool differentiation. Both compose_veo3_prompt and generate_veo3_prompt accept identical parameters and descriptions do not clearly explain when to use each (compose returns 'composed instructions', generate calls Gemini). LLMs will struggle to choose correctly.
Parameter descriptions are minimal and repetitive ('Visual model. One of: ...'). No guidance on format, constraints, or dependencies. Descriptions should explain WHAT each parameter controls and WHY it matters for the prompt.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 67 | 2026-07-28+ | v2 |
No error handling guidance. If CREATIVE_API_KEY is missing or Gemini call fails, LLMs have no recovery path. Error responses should guide the agent to fallback tools or corrective actions.
list_visual_models description does not explain when to call it or what structure it reveals. Should state: 'Call this first to discover available visual models, dialogue modes, music options, output types, aspect ratios, and durations before composing a prompt.'