project-task-router

Project Task Router

Use this skill when a task needs triage before execution. It maps work to the best domain route, skill stack, and model profile using installed skills from:

• nmalinowski/groundwork
• nmalinowski/agents
• nmalinowski/ui-ux-pro-max-skill
• 21st.dev/community/components
• nano-banana-pro

When To Use

• New request is ambiguous or cross-functional.
• You need robust routing across strategy, engineering, design, security, data, and ops.
• You want deterministic, repeatable skill/model selection across tools.

Routing Domains

1. market-discovery-validation
2. product-strategy-positioning
3. architecture-build-planning
4. feature-implementation
5. ui-ux-frontend
6. security-compliance-reliability
7. data-analytics-ai
8. platform-ops-reliability

Routing Method

1. Classify with weighted token/phrase/regex signals.
2. Apply negative-signal penalties.
3. Select top route, plus alternatives when close.
4. Resolve required/optional capabilities to installed skills (dynamic discovery + alias fallback).
5. If confidence is low or route is needs-clarification, ask targeted follow-up questions.

Model Selection Guidance

• deep_reasoning: major ambiguity, high-impact decisions
• balanced: normal planning + execution
• fast_execution: narrow implementation and cleanup work

Router Script

python skills/project-task-router/scripts/route_task.py --task "your task here" --format markdown

Optional installed-skill inputs:

python skills/project-task-router/scripts/route_task.py --task "your task here" --installed-skills "skill-a,skill-b"
python skills/project-task-router/scripts/route_task.py --task "your task here" --installed-skills-file installed-skills.json

Execution Rule

• If task spans multiple domains, sequence discovery/strategy first, then build/design, then assurance.
• If route returns needs-clarification, capture missing intent (deliverable, risk, domain) and rerun.

Subagent Orchestration

1. Route the task.
2. Spawn a review subagent for comprehensive findings.
3. After review completes, spawn a remediation subagent.
4. Remediation policy:
   • Auto-fix low, medium, high
   • Escalate critical findings for explicit approval