AGENT SKILL

salesportl-graphic-design-main-router

Route graphic design services tasks to the smallest reliable skill stack with context-first logic, tool-node awareness, and confidence-gated recursion.

适用职业设计师
使用场景视觉设计
License 许可
MIT
最近更新
2026-11-14
Stars
★ 81
Forks
17

01 前言 Front Matter

name
salesportl-graphic-design-main-router
tags
[ "salesportl", "graphic-design", "niche-suite", "context-loop", "qa-recursive", "tool-node-aware" ]
version
2.0.0
description
Route graphic design services tasks to the smallest reliable skill stack with context-first logic, tool-node awareness, and confidence-gated recursion.

02 资源文件 3

03 SKILL.md 正文

Purpose

Route graphic design services tasks to the smallest reliable skill stack with context-first logic, tool-node awareness, and confidence-gated recursion.

Use when

  • You are operating inside SalesPortl-Graphic-Design-Skills-Suite.
  • You need graphic design services execution quality for graphic design agencies, freelance studios, and in-house brand teams.
  • You want repeatable, improving workflows that learn from each run via context + memory updates.

Industry fit: micro-niche variants

  • brand identity studios
  • social creative studios
  • packaging-focused designers

Standard operating patterns (SOP)

  • creative brief standardization
  • asset review/approval loops
  • brand consistency and revision control

SalesPortl platform tooling focus (if available in workspace)

  • campaign creative packaging for launch sequences
  • click tracking mapping for creative CTAs
  • workflow nodes for concept→review→delivery

Required inputs

  • Business objective and requested deliverable
  • Current context quality (none/partial/verified)
  • Business model, offer mix, and target customer profile
  • Authority boundary for external actions (draft-only unless approved)
  • Available SalesPortl platform features in this workspace

Workflow

  1. Run brand-bootstrap-memory-loop when context is missing, stale, or inconsistent.
  2. Run context-intake to normalize business model, micro-niche, offer mix, and operating constraints.
  3. Classify request into sales, marketing, delivery/operations, client success, or multi-function strategy.
  4. Ask minimum clarifying questions when objective, authority, or context is ambiguous.
  5. Route to 1-3 skills only; include quality-review-orchestrator for all substantive deliverables.
  6. If platform-enabled execution is requested, map the plan to explicit SalesPortl workflow nodes and tool actions.
  7. Launch sub-agents by function for parallel research/execution when complexity is high; merge results through skill-handoff-protocol.
  8. Return selected chain, unresolved unknowns, and confidence score.

Platform execution checklist

  • Campaign orchestration nodes
  • Click-tracking and attribution configuration
  • Workflow/automation nodes
  • Custom agent launch and role-specialized sub-agents
  • CRM pipeline and lead stage automation
  • Calendar, booking, and follow-up automations
  • Website/page/app builder flows (where available)

Output format

- Router chain (skills + order): ...
- Request class and micro-niche fit: ...
- Required SalesPortl features/tool actions: ...
- Missing inputs and clarifying questions: ...
- Sub-agent parallelization plan (if needed): ...
- Confidence score (0-100) and next-step prompt: ...

Confidence scoring rubric (0-5 each)

  • Logic integrity
  • Consistency with context and previous decisions
  • Thoroughness of analysis
  • Completeness for implementation
  • Platform/tooling readiness
  • Niche-specific relevance
  • Target: average >= 4.2 before final handoff

Quality checks

  • Output explicitly reflects this industry plus at least one micro-niche variant.
  • Recommendations map to practical SalesPortl feature usage when requested.
  • Ambiguities are handled by questions, not assumptions.
  • Logic is coherent end-to-end across routing, execution, and QA.
  • Completeness is verified before finalization; partial output is labeled as partial.

Recursive loop control

  1. Evaluate current output against the scoring rubric.
  2. If target score is not met, route revision to the most relevant source skill.
  3. Re-run checks after revision and update confidence score.
  4. Continue until target score is reached or blocking unknowns require user input.

Related skills in this collection

  • salesportl-graphic-design-brand-bootstrap-memory-loop
  • salesportl-graphic-design-context-intake
  • salesportl-graphic-design-offer-service-mapper
  • salesportl-graphic-design-sales-conversion-playbook
  • salesportl-graphic-design-marketing-growth-planner
  • salesportl-graphic-design-delivery-operations-planner
  • salesportl-graphic-design-client-success-expansion
  • salesportl-graphic-design-skill-handoff-protocol
  • salesportl-graphic-design-quality-review-orchestrator

04 同仓库更多 Skills