Agency request intake router

Agency Request Intake Router

Classify any client or internal agency request into the correct operating category, identify the required skill, assess risk level, determine ownership, and produce a structured routing decision with commercial impact assessment.

Why This Exists

Agency operations fail when requests arrive without triage. Briefs get lost, scope creep goes undetected, the wrong team picks up work, and margin erodes silently. This skill acts as the front door — every request gets classified, risk-scored, and routed before any execution begins.

Trigger Conditions

Activate this skill when the user provides any of the following:

• A client request, brief, or campaign ask (in any format — email, Slack message, verbal summary, PDF)
• A CRM, media, data, or reporting request
• A scope change or change order discussion
• A QBR or performance review request
• A vague operational ask that needs routing ("the client wants something")
• An internal request that touches client deliverables
• A question about whether something is in scope or requires a change order

Classification Process

Work through these steps in order. Do not skip steps.

Step 1 — Identify Context

Extract from the request:

• Client / Brand / Tenant: Who is asking? If unclear, ask.
• Requestor: Individual name and role if available.
• Date received: When did this arrive? Default to today if not stated.
• Original channel: Email, Slack, meeting, brief template, etc.
• Urgency signal: Any deadline mentioned, language like "ASAP", "urgent", "for tomorrow".

Step 2 — Classify Request Type

Assign exactly ONE primary type. If the request spans multiple types, pick the dominant one and note secondary types.

Type	Description	Examples
strategy	Strategic planning, positioning, audience definition	Brand strategy workshop, comms planning, competitive analysis
creative	Concept development, copy, design, video production	Campaign concept, banner set, social content calendar
crm	CRM flows, email marketing, automation, segmentation	Journey build, email template, audience segmentation, loyalty program
media	Paid media planning, buying, optimization	Media plan, budget reallocation, new channel activation
data	Analytics, dashboards, tagging, measurement	GA4 setup, dashboard build, attribution model, data layer
reporting	Performance reports, QBRs, post-campaign analysis	Monthly report, QBR deck, campaign wrap-up
production	Adaptation, resizing, trafficking, QA, localization	Banner resizing, ad trafficking, asset adaptation
finance	Billing, invoicing, PO requests, budget reconciliation	Invoice dispute, PO creation, budget tracking
scope-change	Work that modifies the existing SOW or retainer	New deliverable not in contract, headcount increase, platform addition
escalation	Client complaint, service issue, relationship risk	Missed deadline escalation, quality complaint, executive intervention

Step 3 — Scope Assessment

Determine whether this request is covered by the current engagement:

Status	Meaning	Action
in-scope	Clearly within current SOW / retainer	Proceed to routing
unclear	Could be in scope but needs verification against SOW	Flag for account lead review before execution
out-of-scope	Not covered — requires additional budget or approval	Trigger → change-order-generator
requires-change-order	Definitively outside SOW, needs formal CO	Trigger → change-order-generator
requires-leadership	Involves strategic pivots, large budgets, or client escalation	Route to leadership with brief summary

If the user has not shared the SOW or retainer terms, note the assumption and recommend verification.

Step 4 — Risk and Commercial Assessment

Evaluate along these dimensions:

Risk Level (Low / Medium / High / Critical):

• Low: Routine, in-scope, clear ownership, no deadline pressure
• Medium: Tight timeline, cross-functional dependencies, moderate ambiguity
• High: Out-of-scope, involves external vendors, client relationship at stake, margin impact
• Critical: Escalation, legal exposure, executive visibility, revenue at risk

Commercial Impact:

• Estimated hours if quantifiable (or range)
• Whether the work is billable, non-billable, or investment
• Margin risk: does this erode profitability on the account?
• Revenue opportunity: could this grow the engagement?

Step 5 — Route to Next Skill

Based on classification, recommend the next skill in the operating system:

Condition	Recommended Skill
Out of scope or ambiguous scope	scope-audit
Requires formal change order	change-order-generator
Needs capacity check before committing	fte-capacity-sizing
Impacts margin or profitability	margin-simulation
Campaign about to launch, needs QA checklist	campaign-launch-qa
Recurring operational tracking	weekly-control-tower
Leadership needs a summary	executive-growth-memo
Clear and in-scope — just execute	No skill needed; route to owner

If multiple skills apply, list them in priority order with rationale.

Output Format

Produce the routing decision in TWO forms: first as an inline visual artifact (rendered in chat via the Visualizer), then as a structured markdown table below it for copy-paste and downstream consumption.

Visual Artifact (Primary)

Always render the routing decision as an inline HTML widget using the Visualizer. The widget should display:

• A header bar color-coded by risk level: green (Low), amber/yellow (Medium), red (High), dark red/black (Critical)
• The request summary prominently at the top
• A two-column grid showing: Client, Request Type, Scope Status, Risk Level, Commercial Impact, Required Skill, Required Owner, Deadline
• A "Next Action" footer with the recommended action highlighted
• Any ⚠️ TBD fields visually distinct (e.g., dashed border, muted color)
• A clickable "Route to → [skill-name]" button at the bottom that sends the next skill trigger as a chat prompt via sendPrompt()

Use CSS variables for theming so it adapts to light/dark mode. Keep the widget compact — it should feel like a Jira ticket or a triage card, not a full-page report.

Markdown Table (Secondary)

After the visual artifact, also include the routing as a markdown table for reference and downstream use:

## 🔀 ROUTING DECISION

| Field | Value |
|-------|-------|
| **Request Summary** | [1-2 sentence plain-language summary] |
| **Client / Brand** | [Name] |
| **Request Type** | [Primary type from Step 2] |
| **Secondary Types** | [If applicable, otherwise "—"] |
| **Scope Status** | [in-scope / unclear / out-of-scope / requires-change-order / requires-leadership] |
| **Risk Level** | [Low / Medium / High / Critical] |
| **Commercial Impact** | [Brief statement: hours, billable/non-billable, margin effect] |
| **Required Skill** | [Next skill to trigger, or "Direct execution"] |
| **Required Inputs** | [What's needed before the next skill can run] |
| **Required Owner** | [Role or team responsible] |
| **Recommended Next Action** | [Specific, actionable step] |
| **Deadline / Urgency** | [If stated, otherwise "Not specified"] |

When Information Is Missing

If the user's request is too vague to classify confidently, produce a partial routing with the fields you can fill, mark uncertain fields with ⚠️ TBD — [what you need], and ask the user for the missing information. Do not block the entire routing on one missing field.

Examples

Example 1 — Clear in-scope request:

User: "AcmeAuto wants the monthly performance report for April. Due next Friday."

→ Type: reporting · Scope: in-scope · Risk: Low · Skill: Direct execution · Owner: Analytics lead

Example 2 — Scope ambiguity:

User: "The client just asked us to build a loyalty program in Salesforce. We've never discussed this."

→ Type: crm · Scope: out-of-scope · Risk: High · Skill: scope-audit → change-order-generator · Owner: Account Director

Example 3 — Vague ask:

User: "Hey, the client wants 'something for social' by Monday."

→ Type: creative (tentative) · Scope: unclear · Risk: Medium · Missing: brief, format, platform, budget · Action: Request brief or at minimum a 5-minute alignment call before routing further.