Persona Router

Overview

Route a task to the correct Pointline persona and return a copy-paste prompt template the user can run immediately.

When To Use

• •User asks which persona should handle a task.
• •User asks how to switch personas between tasks.
• •User asks for reusable persona prompts.
• •User asks for HFT vs MFT persona variants.

Workflow

1. •If the user explicitly specifies a persona, honor it directly.
2. •Otherwise, classify the task using the routing rules below.
3. •Choose one primary persona. If mixed scope, use a phase split.
4. •Return a ready-to-use prompt with Persona, Task, Focus, and Deliver.
5. •Keep output concise and actionable.

Routing Rules

• •Storage/replay/ingestion semantics: Data Infra Engineer
• •Validation/schema/contract semantics: Data Quality Engineer
• •Research API/feature-pipeline behavior: Research Engineer
• •Signal definition/evaluation and experiments: Quant Researcher
• •Tooling/CI/dev workflow: Platform/DevEx

If Quant Researcher is selected:

• •High-frequency microstructure/replay/latency focus -> Quant Researcher (HFT)
• •Multi-stream seconds/minutes feature workflows -> Quant Researcher (MFT)

Output Format

Always return:

Persona: <role>
Task: <rewritten user task>
Context: <key files/tables/constraints>
Focus:
- <priority 1>
- <priority 2>
- <priority 3>
Deliver:
- <deliverable 1>
- <deliverable 2>

For mixed ownership tasks, return phase split:

Phase 1 Persona: <role>
Phase 1 Task: <task>
Phase 1 Deliver: <output>

Phase 2 Persona: <role>
Phase 2 Task: <task>
Phase 2 Deliver: <output>

Prompt Source

Prefer template language from docs/persona-prompts.md when available in the active repo.

For full variants/examples, load:

• •references/templates.md