Multimodal Vision Tool Routing Matrix
Map multimodal user intents to specific vision APIs, diffusion models, and analyzer tools using an operational routing matrix.
Use this template when designing dispatch logic for agents handling mixed visual and text inputs. It creates a structured routing matrix to match user intent against downstream image tools.
Role: Principal Multimodal Systems Architect specializing in autonomous agent tool orchestration.
Context
- Agent identifier: {{agent_name}}
- Supported input streams: {{input_modalities}}
- Strict latency ceiling: {{latency_sla_ms}}
- Registered visual toolset: {{vision_toolset}}
- Routing prioritization rule: {{routing_policy}}
- Exception fallback strategy: {{error_handling_strategy}}
Task
Synthesize a definitive Multimodal Function Routing Matrix that maps incoming user prompt archetypes to exact tool execution pathways, parameter schemas, and fallback triggers for {{agent_name}}.
Method
- Analyze the functional scope of {{vision_toolset}} against supported {{input_modalities}}.
- Classify incoming query intents into discrete visual task categories including generation, edit, segmentation, and visual interrogation.
- Evaluate execution overhead for each tool against {{latency_sla_ms}} to prioritize lightweight endpoints.
- Apply {{routing_policy}} to resolve ambiguous multimodal instructions requiring chained tool execution.
- Map exact function names, required parameter keys, and deterministic validation requirements for each intent branch.
- Formulate fallback actions and degradation paths following {{error_handling_strategy}} when primary vision tools fail.
- Structure all evaluations into a unified markdown decision matrix with standardized operational columns.
Constraints
- The primary artifact MUST be formatted as a structured markdown matrix table.
- MUST NOT route requests to undocumented or synthetic tool names outside {{vision_toolset}}.
- Every matrix row MUST define expected schema validation keys, fallback triggers, and timeout limits.
- Execution latency per tool combination MUST NOT exceed {{latency_sla_ms}} under nominal operating conditions.
Output format
Provide the deliverable in two sections:
- Routing Matrix Table: A markdown matrix containing columns for Intent Archetype, Input Signal, Primary Tool Call, Payload Signature, Timeout Limit, and Fallback Routine.
- Implementation Notes: Exactly 3 bullet points highlighting parameter validation traps and performance trade-offs.
Self-review
- Verify all tools listed in the matrix exist within {{vision_toolset}}.
- Confirm every intent branch includes a concrete fallback handling rule aligned with {{error_handling_strategy}}.
- Ensure table formatting is valid markdown with no missing column data.
Explicit role, a named task, and discrete steps the model can follow.
Background, inputs and variables the model needs before it starts.
Hard boundaries — what the model must and must not do.
A named, field-level shape for the response.
Ordered work items that force analysis before an answer.
Length and structure that travel across frontier models.
Signal density — instruction weight without padding.
Documented variables so the scaffold adapts to new inputs.
Quality bar, assumptions and behaviour when inputs are thin.
How much real usage the template has behind it.