Mary — Marketo Program Documentation Builder

Use this skill whenever a user wants to document a Marketo program, create a program one-pager, generate program documentation, or audit a program's setup. Triggers include: "document this program", "program one-pager", "program documentation for [name]", "audit this program", "what's in this program", "summarize program [name]", "program overview doc", or any request to produce a structured DOCX describing a Marketo program's configuration, assets, tokens, smart campaign logic, and health status. Always use this skill when the user mentions documenting, auditing, or summarizing a Marketo program — if it involves pulling program details and producing a structured document, this skill applies. Even casual requests like "tell me everything about program X" or "give me a doc on this program" should trigger it.

1. Call marketo_get_programs_by_name(name="<program name>")
2. If multiple results: show the list (name, type, channel, folder) and ask the user to pick one
3. If one result: confirm with the user — "I found [name] ([type], [channel]). Is this the right program?"
4. If no results: try a broader search term, or fall back to get_program_by_name(name="<name>") (direct Marketo MCP, exact match). If still nothing, ask the user to double-check the name.

Gate: Do NOT proceed to Step 2 until you have a confirmed program ID.

Step 2: Pull All Program Data

Read references/api-field-mapping.md now for the full tool chain and field mapping.

Two tool sets are available — allGood wrappers (convenient for search/tokens) and direct Marketo MCP (required for smart campaigns, flow steps, smart list criteria, forms). Use both as needed.

2a. Program metadata

get_program_by_id(id=<programId>)

This returns: name, type, channel, description, status, folder, workspace, tags, costs, SFDC sync info, created/updated dates.

If the allGood wrapper was used in Step 1 and already returned full details, you can skip this call and use that data instead.

2b. My Tokens

get_tokens_by_folder(id=<programId>, folderType="Program")

Returns all local tokens (name, type, value). Inherited tokens are NOT included — note this in the document.

2c. Smart Campaigns

browse_smart_campaigns(folder={id: <programId>, type: "Program"})

This lists all smart campaigns in the program. For each smart campaign returned:

get_smart_campaign_by_id(id=<campaignId>, includeFlowSteps=true)

This returns the campaign metadata AND its flow steps — the actions it performs, in order, including choice steps and wait steps. This is the most valuable data for documentation because flow steps define what the program actually does.

Then get the smart list (triggers + filters) for each campaign:

get_smart_list_by_campaign_id(smartCampaignId=<campaignId>, includeRules=true)

Returns the full filter/trigger criteria with operators, values, and AND/OR logic.

For batch campaigns, also check for scheduled runs:

get_smart_campaign_scheduled_runs(campaignId=<campaignId>)

2d. Emails

browse_emails2(folder={id: <programId>, type: "Program"})

Lists all emails. For each email:

get_email_content(id=<emailId>)

Returns subject line, from name/address, reply-to, content sections, status, and operational flag.

2e. Forms

Use get_folder_content(id=<programId>) to find forms, then for each:

get_form_by_id(id=<formId>)
get_form_fields(id=<formId>)

Returns form metadata and all fields (name, dataType, required, validation rules, progressive profiling settings).

2f. Lists

browse_smart_lists(folder={id: <programId>, type: "Program"})
browse_lists(folder={id: <programId>, type: "Program"})

Returns smart lists and static lists within the program.

2g. Landing Pages and other assets

Use get_folder_content(id=<programId>) to discover any landing pages, reports, or other assets not covered above. Document their names, types, and statuses.

Rate limit awareness

Marketo enforces 100 API calls per 20 seconds and 50,000 calls per day. For programs with many smart campaigns (10+), pace your detail calls. If you hit rate limits, wait 20 seconds and retry.

Summary and confirmation

After pulling all data, print a summary for the user:

Program: [name]
Type: [type] | Channel: [channel] | Status: [status]

Assets found:
  Smart Campaigns: [N] ([X] active, [Y] inactive)
  Emails: [N]
  Landing Pages: [N]
  Forms: [N]
  Smart Lists: [N] | Static Lists: [N]

Tokens: [N] local tokens ([X] populated, [Y] empty/placeholder)

Red flags detected: [N] (will be detailed in the document)

Ready to build the documentation. Proceed?

Wait for user confirmation before building.

Step 3: Analyze Health Flags

Read references/health-checks.md now for the full rule set and detection logic.

Run every health check rule against the collected data. Each rule has an ID, severity (Critical / Warning / Info), condition to check, and a recommendation.

The goal is to surface issues that a Marketo ops practitioner would care about during an audit or handoff — not just data validation, but operational red flags.

Sort findings by severity: Critical first, then Warning, then Info.

If the same rule fires for multiple assets (e.g., three emails are in Draft status), group them into a single finding with a count rather than listing each separately.

If no flags are detected, note "No issues detected" — this is itself valuable information.

Step 4: Build the DOCX

Read references/document-format.md now. Follow the format specification precisely. Also re-check references/api-field-mapping.md for field-to-section mapping.

Use the docx npm library.

Branding

• Header (content pages only — NOT cover page): "Program Documentation" left-aligned, pink #dc4393, 10pt
• Footer (content pages only — NOT cover page): Right-aligned — Mary logo (assets/mary-logo.png, 288 DXA) + "powered by" gray #888888 9pt + allGood logo (assets/allgood-logo.png, 432 DXA). If logos are missing, use text: "Mary powered by allGood" in gray.
• Section headers: Pink #dc4393, Arial 16pt bold

Document Structure — Cover Page + Six Sections

The document starts with a branded cover page, followed by a page break, then the six content sections. See references/document-format.md for the precise cover page spec.

Cover Page (Page 1)

A clean, branded title page with no header/footer. Build it exactly as specified in the Cover Page section of references/document-format.md.

Content to populate from API data:

Insert a page break after the cover page before starting Section 1.

Section 1: Overview

Table type: Key-Value Table (2-column)

Section 2: Configuration

Table type: Key-Value Table (2-column)

Section 3: Asset Inventory

This section has multiple sub-tables for different asset types. Use a sub-heading (bold 13pt, dark gray #666666) before each asset group.

Sub-section: Smart Campaigns Table type: Asset Inventory Table (4-column with pink row numbers)

Sub-section: Emails Table type: Asset Inventory Table (4-column)

Include from name and operational flag in the Details column if relevant.

Sub-section: Landing Pages If any LPs were found, list them with name, URL, and status.

Sub-section: Forms If any forms were found, list them with name, field count, and global/local indicator.

Sub-section: Smart Lists & Static Lists If any lists were found, list them with name and type.

Sub-section: My Tokens Table type: Token Table (3-column)

For rich text tokens, strip HTML tags and show plain text with a "(rich text)" note. Flag tokens with placeholder or empty values using italic gray styling.

Section 4: Program Logic Flow

This is the most valuable section — it documents what the program actually does.

For each smart campaign, create a Smart Campaign Detail Block:

Campaign header: Full-width row with campaign name, type (Trigger/Batch/Executable), and status (Active/Inactive). Pink left border, bold 12pt.

Smart List (Triggers & Filters):

Show the criteria that qualify people for this campaign:

TRIGGERS:
  • [Trigger Name]: [Attribute] [Operator] [Value]

FILTERS:
  1. [Filter Name]: [Attribute] [Operator] [Value]
  2. [Filter Name]: [Attribute] [Operator] [Value]
  Logic: [AND / OR / Advanced: 1 and (2 or 3)]

Use the data from get_smart_list_by_campaign_id with includeRules=true.

Flow Steps:

Show every flow step in order:

FLOW STEPS:
  1. [Action Name] — [Details/Parameters]
     Choice 1: If [condition] → [action]
     Choice 2: If [condition] → [action]
     Default: [action]
  2. Wait — [Duration]
  3. [Action Name] — [Details/Parameters]

Use the data from get_smart_campaign_by_id with includeFlowSteps=true.

Qualification & Schedule:

• Qualification rules: Once / Every time / Once every N days
• Schedule (batch campaigns): recurrence details, next run
• Requestable: Yes/No (for executable campaigns)

If a campaign has no flow steps or no smart list rules, document that fact — it is itself a red flag (HC-13, HC-14).

After all individual campaigns, add a brief Program Flow Summary paragraph that narratively describes: how people enter the program → what happens to them → how they exit or reach success status. This ties the individual campaign details into a coherent story.

Section 5: Health / QA Flags

Table type: Health Flag Table (3-column with severity colors)

Severity cell colors:

• Critical: Red #DC3545 background, white text
• Warning: Orange #F39C12 background, white text
• Info: Blue #4A90D9 background, white text

If no flags were detected, show a single row: "No issues detected — program configuration looks healthy" with a green #28A745 background.

Section 6: Performance Snapshot

Table type: Key-Value Table (2-column)

If membership/performance data is available from the program details response, include:

If performance data is not available in the API response (it often requires separate Lead Database API calls with pagination), show:

"Performance metrics require separate Marketo reporting. Review Program Analyzer or the Email Performance Report in Marketo for engagement data (open rates, click rates, conversions)."

Output

Save to /mnt/user-data/outputs/<program-name>-documentation.docx (sanitize: lowercase, hyphens, no special chars).

Step 5: Validation Checklist

After generating the DOCX, verify every item. Fix any failures before delivering.

Cover page:

• Cover page is Page 1 with NO header/footer
• Mary logo centered near top of page
• "powered by" + allGood logo centered below Mary logo
• Pink horizontal rule (#dc4393, 2pt)
• "PROGRAM DOCUMENTATION" eyebrow label in pink, centered, uppercase
• Program name in large bold type, centered
• 3-cell pill table: Channel | Type | Status — centered, pink text, light gray bg
• Second pink horizontal rule
• Metadata line: Program ID | Workspace | Generated date — centered, gray
• Second metadata line: Folder | Last Modified — centered, gray
• Bottom of page: "For internal use only - Generated by Mary, powered by allGood"
• Page break after cover page before Section 1

Structure:

• All 6 sections present (Overview, Configuration, Asset Inventory, Program Logic Flow, Health/QA Flags, Performance Snapshot)
• Every section has content — no blank sections (use "No data" notes if needed)
• Smart campaign detail blocks present for each campaign found
• Flow steps documented for each campaign (or noted as empty if none exist)
• Smart list criteria documented for each campaign

Data accuracy:

• All API data represented — no silently dropped fields
• Null/missing data shown as "Not set" in italic gray, not left blank
• Token values match API response exactly
• Rich text tokens show plain text with "(rich text)" note
• Dates formatted as MM/DD/YYYY
• Cost values formatted as $X,XXX.XX
• Token values truncated at 200 chars in tables with "..."

Branding & layout:

• Header shows "Program Documentation" left-aligned in pink #dc4393 (content pages only)
• Footer shows right-aligned: Mary logo + "powered by" + allGood logo (content pages only)
• Page margins: 1" top/bottom (1440 DXA), 0.75" left/right (1080 DXA)
• Arial font used throughout
• Table widths are 10,320 DXA
• Section headers are pink #dc4393, 16pt, bold
• Alternating row backgrounds applied (white / #efefef)
• All borders: single, 1pt, #CCCCCC
• Cell margins applied: 100-120 DXA top/bottom, 140 DXA left/right

Health flags:

• Severity colors applied correctly (red Critical / orange Warning / blue Info)
• All detected flags present in Section 5
• Recommendations are actionable and specific

If any check fails, fix it and re-validate before delivering the file.