Add Cloud Flow

Plugin check: Run node "${PLUGIN_ROOT}/scripts/check-version.js" — if it outputs a message, show it to the user before proceeding.

Add Cloud Flow

Connect one or more Power Automate cloud flows to a Power Pages code site, or wire already-registered flows into additional pages/components. For new flows this skill:

• Creates the adx_cloudflowconsumer metadata YAML in .powerpages-site/cloud-flow-consumer/
• Assigns web roles based on the flow's scenario and target audience
• Generates client-side TypeScript/JavaScript service code to trigger the flow with CSRF authentication

For already-registered flows, the skill skips metadata and role creation and goes straight to client-side integration — wiring the existing flow into new UI locations.

Core Principles

• Multiple flows in one run: The user can add several flows at once. All flows are planned together and implemented together before asking for deployment.
• Ask before acting: Present a full HTML plan (all flows, roles, reasoning) before creating any files.
• Web roles drive access: Every flow must have at least one web role. Anonymous Users role is valid but must be confirmed.
• Scenario determines roles: Understand what each flow does and who triggers it before picking roles.
• Use TaskCreate/TaskUpdate: Track all phases upfront before starting any work.

Prerequisites:

• An existing Power Pages code site with .powerpages-site deployed
• PAC CLI authenticated (pac auth who must succeed)
• Azure CLI authenticated (az login --allow-no-subscriptions works whether or not your account has an Azure subscription — Dataverse and Power Platform tokens are AAD-scoped)

Initial request: $ARGUMENTS

Workflow

1. Verify Prerequisites — Locate the project, confirm .powerpages-site exists, inventory web roles and existing flows
2. List Available Flows — Fetch flows from the Power Automate Flow RP API
3. Select Flows & Understand Scenarios — User picks one or more flows; determine scenario and audience for each
4. Determine Web Roles — Propose minimum roles per flow with reasoning
5. Review Plan — Render HTML plan for all flows; get user approval
6. Create Metadata — Write .cloudflowconsumer.yml for each flow; create missing web roles first
7. Client-Side Integration — Generate typed service code to call each flow from the frontend
8. Verify & Summarize — Validate all YAMLs, record skill usage, offer deployment

Phase 1: Verify Prerequisites

Goal: Locate the Power Pages project and confirm all prerequisites are met

Actions:

1. Create todo list with all 8 phases (see Progress Tracking table)

1.1 Locate Project

Look for powerpages.config.json in the current directory or immediate subdirectories.

If not found: Tell the user to create a site first with /create-site.

1.2 Read Config

Read powerpages.config.json to get the siteName (used for display and plan rendering).

1.3 Check .powerpages-site Exists

Look for the .powerpages-site folder in the project root.

If not found:

<!-- gate: add-cloud-flow:1.3.deploy-first | category=plan | cancel-leaves=nothing -->

🚦 Gate (plan · add-cloud-flow:1.3.deploy-first): .powerpages-site missing — cloud flow YAML lives inside it. Deploy first or stop.

Trigger: Phase 1.3 found no .powerpages-site directory. Why we ask: Cloud flow YAML written to a non-existent path will never deploy. Cancel leaves: Nothing — no YAML files written.

Use AskUserQuestion:

If "Yes, deploy now": Invoke /deploy-site first, then continue to Phase 2. If "Cancel": Stop.

1.4 Read Existing Web Roles

Read all .webrole.yml files from .powerpages-site/web-roles/ to inventory available roles. Note each role's id, name, adx_anonymoususersrole, and adx_authenticatedusersrole.

1.5 Check Existing Cloud Flows

Check .powerpages-site/cloud-flow-consumer/ for existing .cloudflowconsumer.yml files. For each file, read:

• processid — the flow's workflowEntityId
• name — the flow's display name
• adx_CloudFlowConsumer_adx_webrole — assigned web role UUIDs

Store these as the already-registered flows list. These flows already have metadata and web roles configured but may need additional frontend integration on other pages or components.

1.6 Detect Frontend Framework

Read package.json to detect the framework (React, Vue, Angular, Astro). Note the framework and its conventions for Phase 7. See ${PLUGIN_ROOT}/references/framework-conventions.md for the detection mapping.

Output: Project root, site name, framework, available web roles, already-registered flows (with processid, name, and web roles)

Phase 2: List Available Flows

Goal: Fetch all Power Automate cloud flows that have a PowerPages trigger

Actions:

Run the list-cloud-flows script (reads environment from pac auth who internally):

node "${PLUGIN_ROOT}/skills/add-cloud-flow/scripts/list-cloud-flows.js"

This calls the Power Automate Flow RP API with the filter properties/definitionSummary/triggers/any(t: t/kind eq 'powerpages'). Results include:

Separate into two lists:

• Unregistered flows: Flows whose id does NOT match any processid from Phase 1.5 — these need full setup (metadata + roles + client-side)
• Already-registered flows: Flows whose id matches a processid from Phase 1.5 — these already have metadata and roles but can be integrated into additional pages/components

Handle errors:

• Exit code 1 with auth message → prompt the user to run az login --allow-no-subscriptions

• Zero flows from API (no flows with a PowerPages trigger exist at all) → tell the user that no Power Automate flows with a PowerPages trigger were found in this environment. Guide them to create a flow first:

  "No Power Automate cloud flows with a PowerPages trigger were found in this environment. To use this skill, you first need to create a flow in Power Automate with the "When a Power Pages flow step is run" trigger, then run this skill again to connect it to your site."

  Stop the workflow — do not continue to Phase 3.

• Zero unregistered flows but already-registered flows exist → do not stop. Continue to Phase 3 with only the already-registered flows available (for additional frontend integration).

Output: List of unregistered flows and list of already-registered flows

Phase 3: Select Flows & Understand Scenarios

Goal: Understand the user's intent, suggest the most relevant flows (including already-registered ones for additional integration), let the user confirm or adjust, and determine the scenario for each selected flow

Actions:

3.1 Analyze User Intent & Suggest Flows

Before showing the full list, analyze the user's initial request ($ARGUMENTS) against both the unregistered flows and the already-registered flows. Match the user's described need to specific flows using display name, description, and scenario inference.

When matching intent, check already-registered flows first — the user may be asking to wire an existing flow into a new page or component, not register a new one.

• If the user described a specific need (e.g., "add automation for the contact form", "connect the email notification flow"): Identify flows whose name or description closely match the request and present them as recommended picks, explaining why each is a good match. Also show the remaining flows as additional options in case the suggestion is wrong.

  Based on your request, I recommend:
    ⭐ 1. Contact Form Submission — Sends an email when a contact form is submitted (Active)
         → Matches your request to add automation for the contact form
  
  Other available flows:
    2. Support Ticket Handler — Creates a case record from user input (Active)
    3. Newsletter Signup — Adds the user's email to a mailing list (Draft)

• If the user's request is generic (e.g., "add a cloud flow", "connect some flows"): Fall back to presenting the full list without recommendations.

  Available flows:
    1. Contact Form Submission — Sends an email when a contact form is submitted (Active)
    2. Support Ticket Handler — Creates a case record from user input (Active)
    3. Newsletter Signup — Adds the user's email to a mailing list (Draft)

Present both categories clearly when both exist:

New flows (not yet registered):
  1. Newsletter Signup — Adds the user's email to a mailing list (Active)

Already registered (available for additional frontend integration):
  2. Contact Form Submission — Already connected, can be wired into more pages
  3. Support Ticket Handler — Already connected, can be wired into more pages

<!-- gate: add-cloud-flow:3.1.select-flows | category=plan | cancel-leaves=nothing -->

🚦 Gate (plan · add-cloud-flow:3.1.select-flows): Multi-select over discovered + already-registered flows. Drives the rest of Phases 4–7.

Trigger: Phase 2 list-cloud-flows returned at least one flow. Why we ask: Wrong flows get registered (new .cloudflowconsumer.yml files written) or wrong existing flows get re-wired into the frontend. Cancel leaves: Nothing — no YAML or client code written yet.

Use AskUserQuestion:

If more than 10 flows are available, ask the user to type names or numbers (comma-separated for multiple).

3.2 Tag Selected Flows

For each selected flow, tag it based on its source:

3.3 Determine Scenario Per Flow

For each selected flow, identify its scenario from the name and description:

<!-- not-a-gate: per-flow scenario clarification — data-gathering for Phase 4 role assignment; no destructive action -->

If a flow's scenario is unclear, use AskUserQuestion per flow:

For integration-only flows, the scenario is still needed to guide where and how the flow is wired into the frontend (Phase 7).

3.4 Determine Client-Side Function Name Per Flow

For each flow, derive a camelCase function name for the client-side service:

• Strip special characters and connector words from the display name
• Example: "PowerPages -> Send an email notification (V3)" → sendEmailNotification
• Example: "Contact Form Submission" → submitContactForm

For integration-only flows, check whether a trigger function already exists in the codebase (from a previous run of this skill). If it does, reuse that function name — do not create a duplicate. Only create a new function if one doesn't exist yet.

Output: Selected flows list, each with tag (new or integration-only), scenario, target audience, and client-side function name

Phase 4: Determine Web Roles

Goal: Propose the minimum set of web roles for each new flow based on its scenario

Skip for integration-only flows — these already have web roles assigned in their .cloudflowconsumer.yml. Do not re-propose or modify roles for them.

Actions:

4.1 Analyze Role Requirements Per Flow (New Flows Only)

Do NOT assign all roles by default. Use the minimum set per flow. Different flows in the same batch can have different role sets.

4.2 Match Against Existing Roles

For each proposed role across all flows, check whether it exists in .powerpages-site/web-roles/. Mark roles as existing or proposed/new.

4.3 Compose Per-Role Reasoning

For each role assigned to each flow, write one sentence explaining why:

• Authenticated Users on a profile form: "Only logged-in users can update their profile."
• Anonymous Users on a contact form: "Unauthenticated visitors must be able to submit a contact request."
• Blog Authors on an email flow: "Blog Authors are the primary actors who submit posts triggering this notification."

4.4 Flag Anonymous Roles

If any flow has Anonymous Users assigned, note this — the HTML plan renders a warning banner and the confirm step must explicitly surface it.

Output: Per-flow role sets with reasoning, list of roles that need creating

Phase 5: Review Plan

Goal: Render the HTML plan for all flows and get user approval before writing any files

Actions:

5.1 Build Plan Data

Assemble the plan JSON (kept in memory — not written to disk). Include all selected flows in CLOUD_FLOWS_DATA, distinguishing new from integration-only:

{
  "SITE_NAME": "<site name>",
  "PLAN_TITLE": "Cloud Flow Integration Plan",
  "SUMMARY": "<N new flows + M integration-only flows, scenarios, key role assignments>",
  "WEB_ROLES_DATA": [
    {
      "id": "<uuid or temp-id>",
      "name": "<role name>",
      "desc": "<description>",
      "builtin": false,
      "isNew": true,
      "isAnonymous": false,
      "color": "#0078d4"
    }
  ],
  "CLOUD_FLOWS_DATA": [
    {
      "flowId": "<workflowEntityId>",
      "name": "<flow display name>",
      "displayName": "<flow display name>",
      "description": "<flow description>",
      "state": "Active",
      "tag": "new | integration-only",
      "scenario": "<scenario type>",
      "rationale": "<why this flow is being added or where it will be additionally integrated>",
      "webRoles": [
        { "id": "<role-uuid>", "reasoning": "<why this role>" }
      ]
    }
  ],
  "RATIONALE_DATA": [
    { "icon": "⚡", "title": "<title>", "desc": "<reasoning>" }
  ]
}

For integration-only flows, webRoles should reflect the existing roles from the .cloudflowconsumer.yml (read-only — not being changed). The rationale should describe where the flow will be additionally integrated (e.g., "Wiring existing Contact Form flow into the support page").

5.2 Render HTML Plan

node "${PLUGIN_ROOT}/scripts/render-cloudflow-plan.js" \
  --output "<PROJECT_ROOT>/docs/cloud-flow-plan.html" \
  --data-inline '<json>'

The render script refuses to overwrite existing files. Before calling it, check if the default output path (<PROJECT_ROOT>/docs/cloud-flow-plan.html) already exists. If it does, choose a new descriptive filename based on context — e.g., cloud-flow-plan-contact-form.html, cloud-flow-plan-apr-2026.html. Pass the chosen name via --output.

Open the rendered file in the default browser (open on macOS, start on Windows, xdg-open on Linux).

5.3 Confirm with User

Give a brief CLI summary: number of flows, scenarios, role count, any anonymous-role warnings.

<!-- gate: add-cloud-flow:5.3.plan-approval | category=plan | cancel-leaves=nothing -->

🚦 Gate (plan · add-cloud-flow:5.3.plan-approval): Final sign-off on the rendered HTML plan before any web role / .cloudflowconsumer.yml / client code is written.

Trigger: Phase 5.2 rendered the HTML plan. Why we ask: Wrong web role assignments committed (especially Anonymous Users on auth-protected flows); orphaned YAML files in .powerpages-site/cloud-flow-consumer/. Cancel leaves: Nothing — no YAML or frontend changes yet.

Use AskUserQuestion:

If "Request changes": Ask what to change, update, re-render, and present again. If "Cancel": Stop.

Output: User-approved plan for all flows

Phase 6: Create Metadata

Goal: Create web roles (if any are new) then write a .cloudflowconsumer.yml for every new flow in the approved plan

Skip for integration-only flows — these already have .cloudflowconsumer.yml and web roles. Do not recreate or modify their metadata. They will be handled in Phase 7 (client-side integration only).

If the approved plan contains only integration-only flows (no new flows), skip this entire phase and proceed to Phase 7.

Actions:

6.1 Create Missing Web Roles

If the approved plan includes roles that do not yet exist (for new flows), invoke /create-webroles now.

After it completes, re-read .powerpages-site/web-roles/ to collect id values for all new roles.

6.2 Process Each New Flow

Repeat steps 6.3–6.4 for every new flow in the approved plan.

6.3 Determine Flow API URL

/_api/cloudflow/v1.0/trigger/<workflowEntityId>

6.4 Run the Create Metadata Script

node "${PLUGIN_ROOT}/skills/add-cloud-flow/scripts/create-cloud-flow-metadata.js" \
  --projectRoot "<PROJECT_ROOT>" \
  --fileSlug "<url-safe-slug>" \
  --flowName "<flow-display-name>" \
  --flowId "<workflowEntityId>" \
  --flowTriggerUrl "" \
  --flowApiUrl "/_api/cloudflow/v1.0/trigger/<workflowEntityId>" \
  --webRoleIds "<uuid1>,<uuid2>" \
  --metadata "<flow-display-name>"

--fileSlug: lowercase display name, special characters replaced with hyphens, max 50 chars. --flowName: display name written as-is into the name YAML field.

The generated YAML (PAC CLI code-site git format):

• adx_* scalar prefix stripped: adx_flowapiurl → flowapiurl, adx_processid → processid
• adx_cloudflowconsumerid → id; adx_websiteid omitted
• M2M key adx_CloudFlowConsumer_adx_webrole kept as-is
• statecode/statuscode defaults omitted; empty strings written as ''
• Fields alphabetically sorted

6.5 Git Commit

After all YAML files (and any new web roles) are created, do a single git commit. Include the HTML plan file (use the actual output path from the render script's JSON response).

Output: One .cloudflowconsumer.yml per new flow, committed (skipped for integration-only flows)

Phase 7: Client-Side Integration

Goal: Generate typed service code that calls each registered flow from the site's frontend

Cloud flows are invoked via an authenticated POST to /_api/cloudflow/v1.0/trigger/<workflowEntityId>. The request must include a CSRF token fetched from /_layout/tokenhtml, and the payload must be wrapped in an eventData key as a JSON-stringified string.

Actions:

7.1 Discover Existing Frontend Patterns

Use the Explore agent to scan the site:

"Check the Power Pages site frontend for existing patterns. Look for:

1. Any existing API service files in src/shared/, src/services/, src/api/, or similar
2. Existing CSRF token handling (__RequestVerificationToken, _layout/tokenhtml)
3. The framework (React / Vue / Angular / Astro) and its conventions (hooks, composables, services)
4. Any existing cloud flow calls (/_api/cloudflow/) already in the codebase — note which flows are already called and from which pages/components Report all findings."

For integration-only flows, the Explore agent's findings in point 4 are critical — they reveal where the flow is already wired in, so the new integration targets a different page/component without duplicating existing call sites.