deploy-pipeline

name	deploy-pipeline
description	Triggers a Power Platform Pipeline deployment run for a Power Pages solution. Selects a target stage, validates the package, optionally configures deployment settings (environment variables, connection references), then deploys and polls for completion. Use when asked to: "deploy pipeline", "run pipeline", "trigger deployment", "deploy to staging", "deploy to production", "run power platform pipeline", "deploy solution via pipeline", "promote solution", "push to staging", "push to production".
user-invocable	true
argument-hint	Optional: stage name or environment label (e.g. 'staging', 'production') to skip stage selection
allowed-tools	Read, Write, Edit, Bash, Glob, Grep, TaskCreate, TaskUpdate, TaskList, AskUserQuestion, mcp__plugin_power-pages_microsoft-learn__microsoft_docs_search, mcp__plugin_power-pages_microsoft-learn__microsoft_docs_fetch
model	opus

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

deploy-pipeline

Triggers a Power Platform Pipeline deployment run. Reads the existing pipeline configuration from docs/alm/last-pipeline.json, selects a target stage, validates the solution package, and deploys it to the target environment.

Prerequisite: Run /power-pages:setup-pipeline first to create the pipeline configuration.

Refer to ${PLUGIN_ROOT}/references/cicd-pipeline-patterns.md for all HAR-confirmed API patterns used in this skill.

Prerequisites

Important: The source (dev) environment must have a Power Platform Pipelines host environment configured. This is set in Power Platform Admin Center (Environments → select env → Pipelines) or via the tenant-level DefaultCustomPipelinesHostEnvForTenant setting. Without this configuration, pac pipeline deploy will fail. The setup-pipeline skill creates the pipeline definition in the host; this admin step connects the dev environment to that host.

• docs/alm/last-pipeline.json exists (created by setup-pipeline)
• .solution-manifest.json exists
• Azure CLI logged in (az account show succeeds)
• PAC CLI logged in (pac env who succeeds)

Phases

Phase 0 — ALM plan gate

plan-alm is the front door. When the user expresses an ALM intent (promote / ship / deploy / move to staging / push to prod / release this version), the orchestrator (/power-pages:plan-alm) should run first. Direct invocation of deploy-pipeline bypasses the orchestrator's pre-plan completeness check, env-var resolution per stage, activation steps, and validation runs. This gate makes that bypass explicit.

Skip rule. If this skill was invoked as part of an active plan-alm orchestration, skip Phase 0 entirely and proceed to Phase 1. The gate helper exposes this via its inExecution block — pass through silently to Phase 1 when:

inExecution.status === "active"

The helper computes this from docs/.alm-plan-data.json — PLAN_STATUS === "In Execution" AND LAST_INVOCATION_AT within the last 60 minutes. check-alm-plan.js refreshes LAST_INVOCATION_AT automatically on every invocation that finds the plan in execution, so each in-chain skill keeps the chain alive for the next one — even multi-hour deploys (deploy-pipeline alone can take 60 min per stage) survive the window without the chain incorrectly de-classifying. Stalled chains (no heartbeat for > 60 min) reclassify as stale-heartbeat and Phase 0 gates fire normally so an abandoned plan doesn't silently bypass user confirmation.

When inExecution.status is anything other than "active" ("not-running", "stale-heartbeat", "no-plan"), run the Phase 0 gate flow below. Branch on the remaining helper fields:

Step 1 — Run the gate helper.

node "${PLUGIN_ROOT}/scripts/lib/check-alm-plan.js" \
  --projectRoot "." \
  --envUrl "{devEnvUrl}" \
  --token "{token}" \
  --solutionId "{solutionId from .solution-manifest.json, if available}"

The helper returns JSON with { exists, stale, staleness: { reason, detail }, generatedAt, planStatus, ... }. The freshness check requires env credentials + solutionId; without those the helper does an existence-only check.

Step 2 — Branch on the result.

Result	Behavior
deferred: true	The user has explicitly deferred ALM for this project (.alm-deferred marker present). Pass through silently to Phase 1 — do not nag.
exists: false	The user hasn't run plan-alm yet. See Step 3.
exists: true, stale: false	Plan is current. Pass through silently to Phase 1.
exists: true, stale: true (reason: solution-modified)	The solution changed after the plan was generated. See Step 4.

Step 3 — No plan. Tell the user:

"No ALM plan exists for this project. /power-pages:plan-alm builds one — it detects the project state, asks about your promotion strategy, and orchestrates this skill in the right order alongside setup-solution / setup-pipeline / activate-site / test-site. Want me to run plan-alm now?"

🚦 Gate (intent · deploy-pipeline:0.no-plan): Fail-closed entry gate when check-alm-plan.js returns exists:false. Helper-script-backed.

AskUserQuestion:

Question	Header	Options
Run /power-pages:plan-alm first?	ALM plan gate	Yes — run /power-pages:plan-alm now (Recommended), Continue without a plan (advanced — I just want to deploy), Cancel

• Yes (Recommended) → invoke /power-pages:plan-alm. It builds the plan and returns — plan-alm is a planner and does not deploy. This skill then re-runs the Phase 0 check (now exists:true) and proceeds to Phase 1.
• Continue without a plan → set BYPASSED_PLAN_GATE = true and proceed to Phase 1. The deploy will still work, but env-var per-stage values, activation, and post-deploy validation aren't orchestrated.
• Cancel → exit cleanly.

Step 4 — Stale plan. Tell the user:

"ALM plan exists from {generatedAt} but the source solution has been modified since (at {solution.modifiedon}). The plan's component count, size analysis, and split decisions may be outdated. Re-running plan-alm will refresh the analysis."

🚦 Gate (intent · deploy-pipeline:0.stale-plan): Fail-closed entry gate when check-alm-plan.js returns stale:true. Helper-script-backed.

AskUserQuestion:

Question	Header	Options
Refresh the plan first?	ALM plan freshness	Refresh — re-run /power-pages:plan-alm (Recommended), Continue with the existing plan, Cancel

• Refresh (Recommended) → invoke /power-pages:plan-alm. After completion, re-run the Phase 0 helper once to confirm freshness; if still stale, surface the detail and proceed to Phase 1 anyway (don't infinite-loop).
• Continue → set STALE_PLAN_ACK = true and proceed to Phase 1.
• Cancel → exit cleanly.

Relationship to Phase 3.5 (pre-deploy completeness check). Phase 3.5 (later in this skill) catches solution gaps right before deploy. Phase 0 catches the bigger miss: the user who never ran the orchestrator at all and is about to push a half-baked deploy through. The two are complementary.

Phase 1 — Verify Prerequisites

Create all tasks upfront at the start of this phase.

Tasks to create:

1. "Verify prerequisites"
2. "Select target stage"
3. "Resolve pipeline info"
4. "Validate package" — in MULTI_RUN_MODE this becomes a single parallel batch (Phase 3.6) covering all N non-skipped solutions; in single-solution / legacy v2 mode it runs per-iteration inline in Phase 4
5. "Configure deployment settings"
6. "Deploy and monitor"
7. "Write deployment record"

Steps:

1. Resolve the project's configured environment URL first, then assert PAC is actually connected to it. verify-alm-prerequisites.js resolves the env from PAC's ambient org context (pac env who), which is not guaranteed to match the project. If PAC has drifted to another environment (duplicate active pac auth profiles, or an external process flipping the active env), the skill would silently run discovery — and later pac pipeline deploy — against the wrong environment (potentially PROD). Assert the match and hard-stop on mismatch rather than pinning --envUrl: pinning would correct only the Dataverse-API calls while later PAC-CLI operations still follow the drifted context, so asserting that PAC itself is on the right env is the safer gate.

   Read the project's recorded env URL (first match wins): .solution-manifest.json → top-level environmentUrl, else powerpages.config.json → environmentUrl. Store as CONFIGURED_ENV_URL. (Both fields are top-level environmentUrl strings; declarative/EDM sites have no powerpages.config.json, so the manifest is the source there.)

   Pass --expectedEnvUrl only when CONFIGURED_ENV_URL actually resolved to a URL. Use the first form when a recorded env URL exists, the second when neither file records one — do not pass an empty or unresolved --expectedEnvUrl "{CONFIGURED_ENV_URL}":

   # CONFIGURED_ENV_URL resolved (recorded in manifest/config) — assert PAC is on it:
   node "${PLUGIN_ROOT}/scripts/lib/verify-alm-prerequisites.js" --require-manifest --expectedEnvUrl "{CONFIGURED_ENV_URL}"
   
   # Neither file records an env URL — omit the flag, fall back to the pac-context default:
   node "${PLUGIN_ROOT}/scripts/lib/verify-alm-prerequisites.js" --require-manifest
   

   --expectedEnvUrl makes the helper compare PAC's resolved env (origin-only) against CONFIGURED_ENV_URL and exit non-zero with an "Environment mismatch: PAC CLI is connected to {X} but this project targets {Y} — run pac env select …" error on mismatch. (As a safety net the helper skips the assertion if the value isn't a parseable env URL — an empty string or an unsubstituted placeholder won't hard-stop — but prefer omitting the flag outright when there's no recorded URL.) Capture output as JSON; extract .envUrl (store as devEnvUrl) and .token (store as DEV_TOKEN). If the script exits non-zero, stop and surface the error — it indicates an env mismatch, or that az login / pac auth / WhoAmI failed.

2. Run detect-project-context.js to read project config and solution manifest:

   node "${PLUGIN_ROOT}/scripts/lib/detect-project-context.js"
   

   Capture output as JSON; extract .solutionManifest (store as solutionManifest), .siteName (store as siteName), and .websiteRecordId. If solutionManifest is null, continue — the manifest is not strictly required at this step (solution info will come from docs/alm/last-pipeline.json).

3. Locate docs/alm/last-pipeline.json — if not found, stop and advise running /power-pages:setup-pipeline first.

   Manifest version check:

   • If schemaVersion === 3, set MULTI_RUN_MODE = true and store deploymentOrder[] as DEPLOYMENT_ORDER. There is a single pipeline with a single set of stages; multi-solution is expressed via N stage runs against the same stage, one per solution in order. This is the current recommended layout.
   • If schemaVersion === 2 (legacy), set MULTI_PIPELINE_MODE = true and store pipelines[] as PIPELINES_LIST. The skill falls back to the older "loop over N separate deploymentpipelines records" behavior. Advise the user to re-run setup-pipeline to migrate to v3.
   • Otherwise read pipelineId, pipelineName, hostEnvUrl, sourceDeploymentEnvironmentId, solutionName, stages[] (single-solution mode — existing behavior).

   In MULTI_RUN_MODE, resolve solutionName + solutionId per iteration of DEPLOYMENT_ORDER. Entries where status === "SkippedEmpty" (typically the {Prefix}_Future buffer) are short-circuited — no stage run is created for them. The single pipelineId / hostEnvUrl / sourceDeploymentEnvironmentId apply to every run.

   In MULTI_PIPELINE_MODE (legacy v2), resolve solutionName per pipeline in the loop (not globally). All pipelines share the same hostEnvUrl and sourceDeploymentEnvironmentId.

4. Acquire host environment token:

   az account get-access-token --resource "{hostEnvOrigin}" --query accessToken -o tsv 2>/dev/null
   

   Where hostEnvOrigin = scheme + host of hostEnvUrl. Store as HOST_TOKEN. If acquisition fails, stop with instructions to check Azure CLI auth.

5. If solutionManifest is available, read solutionManifest.solution.solutionId and solutionManifest.solution.uniqueName from the detected context. Otherwise, use solutionName from docs/alm/last-pipeline.json.

6. Report: "Pipeline: {pipelineName}. Solution: {solutionName}. Available stages: {stage names}."

Phase 1.5 — Ground in current Pipelines deployment documentation

Reference: ${PLUGIN_ROOT}/references/alm-docs-grounding.md

Cap this step at ~30 seconds. If MCP search / fetch errors out, log a one-line note and continue — this skill must remain runnable offline.

1. Run microsoft_docs_search with the query: Power Platform Pipelines stage run validation ValidatePackageAsync DeployPackageAsync approval.
2. Fetch https://learn.microsoft.com/en-us/power-platform/alm/pipelines (and at most one sister page on stage runs, validation, or approval gates) in parallel via microsoft_docs_fetch.
3. Extract a one-paragraph summary of what Microsoft Learn currently says about stage-run lifecycle, validation outcomes, approval-gate workflow, and deploymentsettingsjson overrides. Compare against ${PLUGIN_ROOT}/references/cicd-pipeline-patterns.md and flag any divergence (new status codes, changed stagerunstatus terminal values, new approval-gate API).
4. Use the summary to inform Phase 2+ decisions. Do not silently change skill behavior — surface any divergence to the user as a soft warning before Phase 4 (Create Stage Run + Validate Package).

Phase 2 — Select Target Stage

If the user passed a stage name or environment label as an argument (e.g., staging), match it against stages in docs/alm/last-pipeline.json and skip this question.

🚦 Gate (plan · deploy-pipeline:2.stage): Pick target stage — Staging / Production / etc. Wrong stage selection here is the biggest single failure mode of this skill.

Otherwise, ask via AskUserQuestion:

"Which environment do you want to deploy to? {numbered list of stages from docs/alm/last-pipeline.json, e.g.:

1. Deploy to Staging → {stagingEnvUrl}
2. Deploy to Production → {prodEnvUrl}}"

Store selected stage as SELECTED_STAGE (with stageId, name, targetDeploymentEnvironmentId, targetEnvironmentUrl).

Design rationale — the deploy loop is serial by design. When DEPLOYMENT_ORDER has N entries (e.g. Core → WebAssets → Future for a 2-solution split with a future buffer), the loop runs one solution at a time, in order ascending, halt-on-first-failure. This is intentional. Four constraints stack up and make parallel DeployPackageAsync calls actively harmful, not just non-beneficial:

1. Dataverse import lock at the target env. ImportSolutionAsync takes an env-level lock — only one solution can actively import at a time per environment. Even if the skill fired N DeployPackageAsync calls in parallel, the host would queue them and run them serially anyway. The wall-clock win for parallel deploys is effectively zero.
2. Inter-split dependencies in 3 of 4 split strategies. Change Frequency (Foundation → Integration → Config → Content), Schema (Domain_1..N → Site), and Layer (Core → WebAssets, where WebAssets ppc rows reference the powerpagesite record in Core) all encode strict ordering. Re-ordering breaks the import — a flow that references a table not yet in the target fails with MissingDependency.
3. Per-iteration consent gates. Phase 6.0 (deploy-pipeline:6.0.final-consent) fires before EVERY DeployPackageAsync. The Phase 5 env-var prompt fires per iteration too. Parallel execution would require either batching the gates (explicitly forbidden — see the per-iteration callout below) or running concurrent AskUserQuestions, neither of which the harness supports.
4. Clean failure handling. Serial + halt-on-first-failure means docs/alm/last-deploy.json records per-solution status cleanly. On retry the loop iterates from the start; Dataverse's same-version idempotency turns already-landed solutions into no-ops.

Do not "optimize" this loop by wrapping iterations in Promise.all / Promise.allSettled / await Promise.race. The validation phase (Phase 3.6) IS parallelized — ValidatePackageAsync does NOT take the import lock — but the deploy phase is intentionally serial. If a future Dataverse release removes the env-level import lock, revisit this rationale; until then it is load-bearing, not an oversight.

In MULTI_RUN_MODE (v3 — recommended): The selected stage is looked up once from the single stages[] array. The skill then loops over DEPLOYMENT_ORDER in order, creating one stage run per solution against the same stageId:

1. Phase 3.6 runs first (once, before the loop) — fans out create-stage-run + ValidatePackageAsync + poll-validation-status for every non-skipped solution in parallel. Halts the entire deploy if any solution fails validation. Stores the per-solution stageRunId in VALIDATED_STAGE_RUNS so the serial deploy loop can reuse them.
2. For each entry in DEPLOYMENT_ORDER where status !== "SkippedEmpty": resolve its solutionUniqueName + solutionId, retrieve its stageRunId from VALIDATED_STAGE_RUNS[solutionUniqueName], set ARTIFACT_SOLUTION_NAME / ARTIFACT_SOLUTION_ID / STAGE_RUN_ID, then run Phases 4.4 (fetch deployment notes) → 5 (configure) → 6.0 (consent gate fires every iteration) → 6.1 (deploy) → 6.2 (poll) against the same pipeline. Phase 4.1–4.3 (create stage run + validate + poll-validation) are skipped — Phase 3.6 already did the work in parallel.
3. If any iteration fails (deployment), halt the loop and report which solution failed and which had already landed.
4. Write one docs/alm/last-deploy.json at the end summarizing all runs for the selected stage. Record per-solution status (Succeeded / Failed / NotAttempted / SkippedEmpty) plus the shared pipelineId.

⚠ Per-iteration gate firing — non-negotiable. Inside the loop, the full Phase 3 → 3.5 → 4 → 5 → 6.0 → 6.1 → 6.2 → 7 sequence runs for each solution. Do NOT batch validation across solutions, do NOT batch the Phase 6.0 consent gate, and do NOT treat any upstream answer (Phase 2 stage selection, --stage argument, the previous iteration's "Deploy now") as covering subsequent iterations. The Phase 6.0 gate fires N times for N non-skipped solutions in DEPLOYMENT_ORDER. If you find yourself proceeding from iteration 1's success directly to iteration 2's DeployPackageAsync without a fresh Phase 6.0 prompt, you have skipped the gate.

In MULTI_PIPELINE_MODE (v2 — legacy): The selected stage label (e.g., "Staging") is matched against each pipeline's stages[] — each pipeline has its own stageId for the same target environment. All subsequent phases (validate, deploy, poll) are looped over PIPELINES_LIST in order:

1. Loop iteration i: use pipelines[i].stageId where stage label matches SELECTED_STAGE.name, pipelines[i].solutionName, etc.
2. If any iteration fails (validation or deployment), halt the loop and report which pipeline failed and which were already deployed.
3. Write one docs/alm/last-deploy.json at the end summarizing all pipeline runs for this stage. Record per-pipeline status (Succeeded / Failed / NotAttempted) so a retry can tell which ones still need to run.

⚠ Per-iteration gate firing also applies here. Same rule as MULTI_RUN_MODE: each pipeline in the loop gets its own Phase 4 / 5 / 6.0 / 6.1 / 6.2 sequence. The Phase 6.0 consent gate fires once per pipeline. Do NOT batch.

Partial-deploy risk. When the loop halts (e.g., Core succeeded, WebAssets failed), the target environment is left in a mixed state — there is no automatic rollback of solutions that already imported. The per-solution (v3) or per-pipeline (v2) status in docs/alm/last-deploy.json is the source of truth for what landed. When the user re-runs deploy-pipeline after fixing the failure, the loop iterates all entries again from the start; rely on the solution-import idempotency (same version = no-op) rather than skipping. Warn the user of this before starting a multi-solution deploy to production.

Check docs/alm/last-deploy.json — if the last deployment to this stage failed, warn the user:

"The last deployment to {stageName} had status: Failed. Would you like to retry? 1. Yes, retry / 2. No, cancel"

Phase 2.5 — Pre-flight: target env blocked-attachments check

Power Pages code-site solutions almost always contain .js bundle chunks (Vite/Rollup output) as Web File components. If the target env's blockedattachments setting includes .js, ImportSolutionAsync will reject every web-file write — typically 50-75 minutes into an import for sites with thousands of bundle chunks (real-world: a Content solution failed at 3,909 rejected .js files on Staging after the same issue had already been fixed on Dev). The reactive Phase 7.6 handler will detect this and offer unblock-and-retry, but the user has already burned an hour. This pre-flight catches it in 10 seconds.

Skip rule. This check is for Power Pages projects only. Skip when powerpages.config.json has no websiteRecordId (non-Power-Pages ALM run — pure data-model solution, etc.). Skip when the user's plan/manifest indicates no solution being deployed has Web File componentType (rare in code-site projects but possible for back-end-only solutions).

Detection signal. In MULTI_RUN_MODE / MULTI_PIPELINE_MODE: read .solution-manifest.json and check whether any entry in solutions[] has componentTypes including "Web File". In single-solution mode: assume true for any Power Pages project (the umbrella solution carries web files).

Steps:

1. Switch PAC CLI context to the target environment so fix-blocked-attachments.js queries the right env:

   pac env select --environment "{SELECTED_STAGE.targetEnvironmentUrl}"
   

2. Run the helper in dry-run mode to detect the current state:

   node "${PLUGIN_ROOT}/scripts/lib/fix-blocked-attachments.js" \
     --envUrl "{SELECTED_STAGE.targetEnvironmentUrl}" \
     --extensions js,css \
     --dry-run
   

3. Capture the output as JSON. Inspect wasBlocked[]:

   • wasBlocked: [] → target env doesn't block the relevant extensions. Switch PAC CLI back to source (pac env select --environment "{sourceEnvUrl}") and proceed to Phase 3. No prompt, no noise.

   🚦 Gate (consent · deploy-pipeline:2.5.blocked-attachments): Pre-flight — modify target env's blockedattachments security setting (tenant-wide impact). Reversible from PPAC. Skipping costs 50–75 min of wasted import.

   • wasBlocked: ["js"] or includes other media-relevant extensions → the deployment WILL fail mid-import. Prompt the user immediately via AskUserQuestion (do NOT bury this in chat — it MUST gate Phase 3 progression):

     Pre-flight detected an issue. The target environment {targetEnvName} currently blocks file types that this solution needs: {wasBlocked.join(', ')}. Power Pages code sites ship .js bundle chunks as web files — if you proceed without unblocking, the deployment will run for ~50-75 minutes and then fail (the failure is recoverable via Phase 7.6's retry path, but the wasted time is not). The block can be removed in 10 seconds.

     Note: this modifies an environment-level security setting that affects all users of {targetEnvName}. Reversible from PPAC → Environments → {targetEnvName} → Settings → Product → Features → Blocked Attachments.

     Question	Header	Options
     Allow removing the block on {wasBlocked.join(', ')} for the {targetEnvName} environment so the deployment can proceed?	Unblock attachments	Yes — unblock these types and continue, No — proceed anyway (Phase 7.6 will catch the failure after deploy and prompt again), Cancel deploy

4. Branch on the answer:

   • Yes — unblock and continue: invoke fix-blocked-attachments.js without --dry-run (same --extensions). Read the result and confirm changed: true + removed[] is non-empty. Switch PAC CLI back to source (pac env select --environment "{sourceEnvUrl}"). Proceed to Phase 3. Record the unblock action in the eventual docs/alm/last-deploy.json preflightActions[] block so the deploy summary has an audit trail.

   • No — proceed anyway: leave the setting unchanged. Switch PAC CLI back to source. Proceed to Phase 3. Tell the user clearly: "Continuing without unblocking. If the deployment fails on AttachmentBlocked, Phase 7.6 will offer the same unblock prompt — but you'll have spent ~50-75 minutes getting there."

   • Cancel deploy: stop cleanly. Do not modify any environment setting. Do not create the stage run. Tell the user how to re-invoke when ready.

5. Always switch PAC CLI back to the source environment before exiting this phase. Subsequent phases assume PAC points at the source unless they explicitly switch to the target.

Why pre-flight + reactive both exist. The reactive Phase 7.6 handler stays in place because (a) the pre-flight only inspects the env-level blockedattachments setting — there are other rare blocked-attachment causes (per-table file column attachment policies) that only surface during the actual import; (b) the env's blocklist could change between pre-flight and import (rare but possible if another admin edits it concurrently); (c) backward compatibility with existing flows where the user invoked deploy-pipeline directly and skipped Phase 2.5 via legacy SKILL.md. The two paths are complementary, not redundant.

Phase 3 — Resolve Pipeline Info

Call RetrieveDeploymentPipelineInfo to get the authoritative source environment ID and available solution artifacts:

GET {hostEnvUrl}/api/data/v9.1/RetrieveDeploymentPipelineInfo(DeploymentPipelineId={pipelineId},SourceEnvironmentId='{BAP_SOURCE_ENV_ID}',ArtifactName='{solutionName}')
Authorization: Bearer {HOST_TOKEN}
OData-MaxVersion: 4.0
OData-Version: 4.0
Accept: application/json

Where BAP_SOURCE_ENV_ID = the BAP GUID of the dev environment (from pac env list, stored in docs/alm/last-pipeline.json or available from pac env who).

Extract:

• SourceDeploymentEnvironmentId — use as the devdeploymentenvironment binding in the stage run. Store as sourceDeploymentEnvironmentId.
• StageRunsDetails[].DeploymentStage — confirms available stages and their IDs
• EnableAIDeploymentNotes — store as AI_NOTES_ENABLED (bool)

Use solutionId from .solution-manifest.json as ARTIFACT_SOLUTION_ID and uniqueName as ARTIFACT_SOLUTION_NAME.

If RetrieveDeploymentPipelineInfo returns 404 (older Pipelines package): use the navigation property to find the source deployment environment:

GET {hostEnvUrl}/api/data/v9.1/deploymentpipelines({pipelineId})/deploymentpipeline_deploymentenvironment?$select=deploymentenvironmentid,name,environmenttype

Filter for environmenttype = 200000000 to get the source record. Use deploymentenvironmentid as the sourceDeploymentEnvironmentId. For the artifact/solution list, use sourceDeploymentEnvironmentId from docs/alm/last-pipeline.json and solutionName from .solution-manifest.json as fallbacks. Set a flag VALIDATE_PACKAGE_UNAVAILABLE = true to skip Phase 4.2–4.3 and use the PAC CLI path in Phase 6.

If RetrieveDeploymentPipelineInfo returns a NON-404 error (e.g. 400/4xx/5xx) — observed: some Pipelines packages return 400 for this call even though ValidatePackageAsync works fine — do NOT set VALIDATE_PACKAGE_UNAVAILABLE. The 404 branch above is specifically for older packages that lack the OData validation API; a 400 is just this metadata call failing, not the validation API being absent. Instead, fall back to sourceDeploymentEnvironmentId from docs/alm/last-pipeline.json (and solutionName from .solution-manifest.json) and continue the normal ValidatePackageAsync flow (Phase 4 onward). If docs/alm/last-pipeline.json is somehow missing sourceDeploymentEnvironmentId, use the same deploymentpipeline_deploymentenvironment navigation-property GET shown in the 404 branch above to recover it (still without setting VALIDATE_PACKAGE_UNAVAILABLE). Only a genuine 404 — or a later ValidatePackageAsync 404 (Phase 4.2) — routes to the PAC-CLI path.

Phase 3.5 — Pre-deploy Completeness Check

A pipeline's ValidatePackageAsync confirms the solution zip is importable on the target, but it does not tell you whether the solution zip itself covers every component that exists on the source site. Components added after setup-solution last ran (server logic, cloud flows, bots, env vars, etc.) can be silently left behind.

Run the shared site-inventory helper against the source (dev) environment:

node "${PLUGIN_ROOT}/scripts/lib/discover-site-components.js" \
  --envUrl "{devEnvUrl}" --token "{DEV_TOKEN}" \
  --siteId "{websiteRecordId from .solution-manifest.json}" \
  --publisherPrefix "{publisherPrefix from .solution-manifest.json}" \
  --solutionId "{solutionId from .solution-manifest.json}" \
  --projectRoot "."

Parse stdout and evaluate missing.*. Before doing anything else, capture the pre-sync state so a post-sync re-confirmation gate can show what changed:

PRE_SYNC_VERSION = solutionManifest.solution.version   // from .solution-manifest.json read in Phase 1
PRE_SYNC_MISSING = { siteComponents, siteLanguages, cloudFlows, envVarDefinitions, customTables, ... }   // from the discovery stdout above

Then:

• All missing.* empty → proceed to Phase 4.

🚦 Gate (progress · deploy-pipeline:3.5.completeness): Source solution incomplete vs live site. Sync first, deploy anyway (gap ships), or cancel.

• Any non-empty → report a short summary ("Solution is missing {N} components"). Ask via AskUserQuestion:

  "The source solution appears incomplete relative to the live site. What would you like to do?

  1. Run /power-pages:setup-solution now (sync mode) — adopts missing components and bumps the version, then re-confirm with you before deploying (Recommended)
  2. Deploy anyway — the missing components will not reach the target
  3. Cancel — I'll investigate first"

  • Option 1 — Sync first, then re-confirm before deploy:

    1. Invoke /power-pages:setup-solution (auto-detects the existing manifest, enters sync mode, adopts missing components, bumps the version). Wait for completion. setup-solution's final refresh step writes LAST_SYNC_AT into docs/.alm-plan-data.json so subsequent check-alm-plan.js calls do NOT falsely flag the plan as stale just because the sync bumped solutions.modifiedon past GENERATED_AT — the freshness reference becomes max(GENERATED_AT, LAST_SYNC_AT).
    2. Re-read .solution-manifest.json and capture POST_SYNC_VERSION = solutionManifest.solution.version.
    3. Re-run the discovery helper. If any missing.* remain non-empty, repeat the Phase 3.5 prompt above.
    4. Otherwise compute NEWLY_ADOPTED as a per-category set difference between PRE_SYNC_MISSING and the second discovery run's missing.* (the items that disappeared are what setup-solution just adopted into the solution). Total count = sum of all category lengths.

    🚦 Gate (progress · deploy-pipeline:3.5.post-sync): Post-sync re-confirm. Solution version bumped + components adopted — user inspects delta before deploy proceeds.

    5. Re-confirm with the user before proceeding to Phase 4 — the solution about to ship is now different from what the user originally saw when they started the deploy. Use AskUserQuestion:

       "Sync complete.

       {solutionUniqueName} is now v{POST_SYNC_VERSION} (was v{PRE_SYNC_VERSION}) with {NEWLY_ADOPTED.total} newly-adopted components:

       • {first 3-5 names by category — prefer high-signal categories: cloud flows, server logic, env var definitions, then site components}
       • {if more remain: + {N} more across {category list}}

       About to deploy this updated solution to {SELECTED_STAGE.name} ({targetEnvUrl}).

       Continue with the deployment?"

       Question	Header	Options
       Continue with the deployment?	Post-sync approval	Yes — deploy v{POST_SYNC_VERSION} to {SELECTED_STAGE.name} (Recommended), Pause — I want to review the new solution contents first, Cancel — abort the deploy

       • Yes → proceed to Phase 4 with the post-sync solution.
       • Pause → exit deploy-pipeline cleanly with a short note ("Paused after sync. Re-run /power-pages:deploy-pipeline when you're ready to ship v{POST_SYNC_VERSION} to {SELECTED_STAGE.name}.") so the user can inspect the synced manifest / Dataverse state and resume manually. Do not write docs/alm/last-deploy.json — no deployment happened. Skip the skill-tracking call too.
       • Cancel → stop the skill. Same no-marker / no-tracking rule applies.

  • Option 2 — record the deliberate gap in docs/alm/last-deploy.json under a knownGaps field so the audit trail is preserved.

  • Option 3 — stop.

Why the post-sync gate exists: this skill is the gate that promotes a solution to staging or production — the moment of staging promotion is the last place to catch surprises. When sync mode runs mid-deploy, it produces a different solution version than the one the user had in mind when they invoked the skill. Re-confirming after sync gives the user an explicit chance to inspect the version bump and the list of newly-adopted components before they reach the target environment. The Phase 3.5 trigger is intentional; the post-sync re-confirmation is the safety on top of it. Same principle applies when this skill is invoked from plan-alm orchestration — plan-alm's plan approval (Phase 4) covers the pre-sync state; this gate covers the delta introduced by mid-deploy sync.

Why Phase 3.5 exists in the first place: the ALM-aware-by-default rule in AGENTS.md requires the completeness check at every gate where a solution leaves its source environment.

Phase 3.6 — Parallel Validation Batch (MULTI_RUN_MODE only)

Skip this entire phase when MULTI_RUN_MODE = false (single-solution mode, or legacy MULTI_PIPELINE_MODE v2). Single-solution and v2 each create + validate exactly one stage run inside Phase 4 below — there's nothing to parallelize. Resume at Phase 4.

This phase compresses the per-solution create-stage-run → ValidatePackageAsync → poll-validation-status chain by running all N solutions concurrently. ValidatePackageAsync does not acquire the env-level import lock that pins DeployPackageAsync to serial execution — it's a structural check on the solution package and per-stage-run state, so the platform happily processes N validations in parallel. For a typical 5-solution split with 60–180s validations, this drops the validation phase from N × ~120s (≈10 min) to roughly the slowest single validation (≈3 min).

The deploy phase (6.1 / 6.2) remains strictly serial — see the Design rationale callout in Phase 2.

3.6.1 Build the input file.

Filter DEPLOYMENT_ORDER down to entries that need validation:

• Include: any entry whose status !== "SkippedEmpty" AND isFutureBuffer !== true.
• Exclude: SkippedEmpty and isFutureBuffer: true entries — the Future buffer is a reserved 0-component placeholder and there's nothing to validate.

Resolve each remaining entry's solutionId from .solution-manifest.json (schemaVersion: 2 solutions[]) if not already on the deployment-order entry. Write to a tmp file:

node -e "require('fs').writeFileSync('./docs/alm/.validation-batch.json', JSON.stringify({{VALIDATION_SPECS}}))"

Where {{VALIDATION_SPECS}} is the array [{ solutionUniqueName, solutionId }, …].

3.6.2 Run the batch validator.

node "${PLUGIN_ROOT}/scripts/lib/validate-stage-runs-batch.js" \
  --hostEnvUrl "{hostEnvUrl}" \
  --token "{HOST_TOKEN}" \
  --pipelineId "{pipelineId}" \
  --stageId "{SELECTED_STAGE.stageId}" \
  --sourceDeploymentEnvironmentId "{sourceDeploymentEnvironmentId}" \
  --solutionsFile ./docs/alm/.validation-batch.json

Capture stdout as JSON: const batch = JSON.parse(output). Delete the tmp file (./docs/alm/.validation-batch.json) — it's transient. Build VALIDATED_STAGE_RUNS = { [solutionUniqueName]: { stageRunId, validationResults } } from batch.results.

If VALIDATE_PACKAGE_UNAVAILABLE propagates (any per-solution error matches ValidatePackageAsync not available on this Pipelines package): set VALIDATE_PACKAGE_UNAVAILABLE = true globally, skip the rest of Phase 3.6 (the stage runs created so far stay around — they're harmless validated-but-not-deployed records), and proceed to Phase 4 in single-solution-fallback shape, which routes each iteration through the pac pipeline deploy CLI fallback in Phase 6. This is the same code path the older Pipelines package versions take.

3.6.3 Branch on the batch outcome.

batch.allPassed	batch.pendingApproval	batch.failed + batch.timedOut	Behavior
true	0	0	All validations passed. Report a single line: "Validated {N} solution(s) in parallel — all passed." Proceed to Phase 4.
false	> 0	0	One or more validations are awaiting approval. See 3.6.4 (Pending Approval batch handling) below.
false	—	> 0	One or more validations failed or timed out. See 3.6.5 (Halt on batch validation failure) below.

3.6.4 Pending Approval batch handling.

🚦 Gate (pause · deploy-pipeline:3.6.batch-pending-approval): External wait — one or more solutions hit stagerunstatus=200000005 (Pending Approval) during parallel validation. User approves all in PPAC, then we re-poll. Cancel leaves N validated-but-pending stage runs on the host (the user can either approve them later and re-invoke, or cancel them in PPAC). Fires once per batch — not once per pending solution.

Surface the affected solutions in a single message (not per-solution) and pause. Use AskUserQuestion:

"Validation for {batch.pendingApproval} of {batch.total} solution(s) is awaiting approval before it can complete: {bulleted list of result.solutionUniqueName where status === 'PendingApproval'}

Approve all of them in Power Platform: make.powerapps.com → Solutions → Pipelines → for each stage run listed above → Approve. Then return here.

The other {batch.succeeded} solution(s) already validated successfully — they'll deploy after you approve."

Question	Header	Options
Approvals complete?	Batch validation approval	Yes — I approved all of them; re-poll, No — cancel the deploy

• Yes: re-run validate-stage-runs-batch.js in --rePoll mode. The helper accepts existing stage run IDs and skips the create-stage-run + ValidatePackageAsync calls — it just runs the poll-and-probe pattern against each stageRunId. After user approval, the stage run transitions 200000005 (PendingApproval) → 200000006 (Validating) → 200000007 (ValidationSucceeded); the helper's probe correctly distinguishes "still pending" (the user clicked Yes prematurely) from "real timeout" so the agent doesn't have to interpret a generic timeout error.

  Build a tmp file with only the previously-pending entries (carry stageRunId from the original batch result):

  node -e "require('fs').writeFileSync('./docs/alm/.repoll-batch.json', JSON.stringify({{PENDING_SPECS_WITH_STAGERUNIDS}}))"
  node "${PLUGIN_ROOT}/scripts/lib/validate-stage-runs-batch.js" \
    --hostEnvUrl "{hostEnvUrl}" \
    --token "{HOST_TOKEN}" \
    --rePoll \
    --solutionsFile ./docs/alm/.repoll-batch.json
  

  Where {{PENDING_SPECS_WITH_STAGERUNIDS}} is the array [{ solutionUniqueName, solutionId, stageRunId }, …] filtered to entries with status === 'PendingApproval' from the original batch.

  Capture stdout as JSON; delete the tmp file (./docs/alm/.repoll-batch.json). Merge the updated outcomes into VALIDATED_STAGE_RUNS keyed by solutionUniqueName. Then branch on the rePoll batch's tally:

  • All succeeded → proceed to Phase 4 with the full set of validated stage runs.
  • One or more still PendingApproval → fire the same gate again (the approval hadn't propagated; the user gets a fresh "approve in PPAC, then re-poll" prompt). Loop until either all approved or the user cancels.
  • One or more Failed / Timeout / Error → fall through to 3.6.5 (treat as batch validation failure).

• No: stop cleanly. The validated stage runs and the pending stage runs remain on the host; re-invoking the skill picks up where this left off (the user can approve and re-run).

3.6.5 Halt on batch validation failure.

🚦 Gate (plan · deploy-pipeline:3.6.batch-validation-failed): One or more solutions failed validation in the parallel batch. Surface per-solution validationResults for the failing entries, and let the user decide whether to abort the entire deploy or proceed with only the succeeded solutions (advanced — leaves a known dependency gap on the target). Cancel leaves N validated stage runs on the host; the user can clean them up in PPAC or re-invoke after fixing the source.

Surface the failing entries with their validationResults (which is the double-encoded JSON string from the Dataverse validationresults field — JSON.parse it twice when displaying to extract SolutionValidationResults[].Message). Use AskUserQuestion:

"{batch.failed + batch.timedOut} of {batch.total} solution(s) failed parallel validation:

{for each failing result: a short block with solutionUniqueName, status, top Message from validationResults}

The other {batch.succeeded} solution(s) passed and have validated stage runs ready to deploy. Deploying only the succeeded subset risks leaving a dependency gap on the target — e.g. shipping _Content without its prerequisite _Foundation produces broken site behavior. Recommended: abort, fix the source, re-invoke.

What would you like to do?"

Question	Header	Options
Next step?	Batch validation failed	Abort the entire deploy (Recommended), Deploy only the succeeded subset (advanced — accept the gap), Cancel and investigate

• Abort / Cancel: stop cleanly. Write a minimal docs/alm/last-deploy.json with status: "ValidationFailed" per failing solution and status: "NotAttempted" for the rest, so the next invocation can see what happened. Do not call DeployPackageAsync.
• Deploy only the succeeded subset: set DEPLOY_SUBSET_ACK = true and filter DEPLOYMENT_ORDER down to just the entries with status === 'Succeeded' in VALIDATED_STAGE_RUNS. Record the skipped entries in docs/alm/last-deploy.json knownGaps. Proceed to Phase 4. Never offer this option as the default — it's a foot-gun and the recommended path is always to abort.

3.6.6 Persist the batch summary.

Append a batchValidation object to the in-memory state that will be written to docs/alm/last-deploy.json in Phase 7. Source most fields directly from the batch JSON returned by validate-stage-runs-batch.js:

{
  "totalSolutions": <batch.total>,
  "succeeded": <batch.succeeded>,
  "failed": <batch.failed>,
  "pendingApproval": <batch.pendingApproval>,
  "timedOut": <batch.timedOut>,
  "elapsedSeconds": <batch.elapsedSeconds>,
  "perSolutionStageRunIds": { "<solutionUniqueName>": "<stageRunId>", ... }
}

Build perSolutionStageRunIds by reducing batch.results into { [r.solutionUniqueName]: r.stageRunId } for every entry where stageRunId is non-null (including failed-but-stage-run-was-created entries — preserves the deep-link to PPAC for debugging). elapsedSeconds comes from the helper directly (it wall-clock-measures the fan-out internally — no need to wrap the bash invocation in a timer).

This gives the next invocation (and any audit consumer) a record of the parallel-validation outcome distinct from the serial deploy outcomes. refresh-alm-plan-data.js's deploy-pipeline phase ingests this block into planData.pipelineMeta.lastDeploy.batchValidation so the rendered ALM plan can show the parallel-validation timing.

Phase 4 — Create Stage Run + Validate Package

In MULTI_RUN_MODE, Phase 3.6 already created the stage run + ran validation for the current iteration's solution in parallel with its siblings. Inside the per-iteration loop, skip Steps 4.1–4.3 entirely: retrieve STAGE_RUN_ID from VALIDATED_STAGE_RUNS[solutionUniqueName].stageRunId and validationResults from the same record. Jump directly to Step 4.4 (fetch deployment notes) and then Phase 5.

In single-solution mode and legacy MULTI_PIPELINE_MODE (v2), run Steps 4.1–4.4 inline as documented below.

Use Node.js https module for all Dataverse calls (curl has encoding issues on Windows).

4.1 Create stage run using create-stage-run.js:

node "${PLUGIN_ROOT}/scripts/lib/create-stage-run.js" \
  --hostEnvUrl "{hostEnvUrl}" \
  --token "{HOST_TOKEN}" \
  --pipelineId "{pipelineId}" \
  --stageId "{SELECTED_STAGE.stageId}" \
  --sourceDeploymentEnvironmentId "{sourceDeploymentEnvironmentId}" \
  --solutionId "{ARTIFACT_SOLUTION_ID}" \
  --artifactName "{ARTIFACT_SOLUTION_NAME}"

Capture stdout as JSON: const result = JSON.parse(output). Extract result.stageRunId and store as STAGE_RUN_ID.

If the script exits non-zero, surface the error — likely a pipeline configuration issue (400) or a conflict (409). Both include the Dataverse error body in the message.

Note on field bindings: The script uses the v9.2 API and msdyn_ prefixed nav properties ([e-mail], [e-mail], [e-mail]). These are the HAR-confirmed names for the current Pipelines package. Older package versions used different field names (e.g., deploymentstageid); the script handles both 201 (JSON body) and 204 (OData-EntityId header) response codes.

Store as STAGE_RUN_ID.

4.2 Trigger package validation (returns 204 — not 200):

POST {hostEnvUrl}/api/data/v9.0/ValidatePackageAsync
Content-Type: application/json
Authorization: Bearer {HOST_TOKEN}

{"StageRunId": "{STAGE_RUN_ID}"}

Treat HTTP 204 as success.

If ValidatePackageAsync returns 404: this Pipelines package version doesn't support the direct OData validation API. Set VALIDATE_PACKAGE_UNAVAILABLE = true. Skip Phase 4.2–4.3 and proceed directly to Phase 5 (deployment settings), then use the pac pipeline deploy CLI fallback in Phase 6.

4.3 Poll validation using poll-validation-status.js:

node "${PLUGIN_ROOT}/scripts/lib/poll-validation-status.js" \
  --hostEnvUrl "{hostEnvUrl}" \
  --token "{HOST_TOKEN}" \
  --stageRunId "{STAGE_RUN_ID}" \
  --intervalMs 5000 \
  --maxAttempts 36

Capture stdout as JSON: const result = JSON.parse(output). On non-zero exit, the error message will include validation failure details or a timeout message.

The script polls msdyn_operation on the stage run record. While it equals 200000201 the validation is still in progress; once it changes the script checks msdyn_stagerunstatus — if 200000003 (Failed) it throws with msdyn_validationresults; otherwise it returns { stageRunId, validationResults, stageRunStatus }.

Terminal validation values:

• stageRunStatus 200000007 (Validation Succeeded) → proceed to Phase 5
• Script throws on 200000003 (Failed) — stop, display the validationresults from the error message
• Script throws on timeout — stop with the message

Important: validationresults is a double-encoded JSON string — call JSON.parse() on it twice to get the object. The object has shape: { ValidationStatus, SolutionValidationResults: [{ SolutionValidationResultType, Message, ErrorCode }], SolutionDetails, MissingDependencies }.

Surface any SolutionValidationResults entries to the user as warnings. Pay special attention to:

• ErrorCode: -2147188672 — managed/unmanaged conflict: "The solution is already installed as unmanaged but this package is managed." The user must uninstall the existing solution from the target environment first, then retry.
• Missing connection references or environment variable gaps

🚦 Gate (pause · deploy-pipeline:4.pending-approval): External wait — PP Pipelines stagerunstatus=200000005 (Pending Approval). User must approve in PPAC. Cancel leaves the run pending on the host. In MULTI_RUN_MODE / MULTI_PIPELINE_MODE this gate fires per loop iteration that hits Pending Approval — not once per skill invocation.

If stageRunStatus = 200000005 (Pending Approval): inform the user they need to approve in Power Platform (make.powerapps.com → Solutions → Pipelines → find this run → Approve). Ask via AskUserQuestion: "Have you approved the validation? 1. Yes, continue / 2. No, cancel"

4.4 Fetch AI-generated deployment notes (if AI_NOTES_ENABLED = true):

GET {hostEnvUrl}/api/data/v9.0/deploymentstageruns({STAGE_RUN_ID})?$select=aigenerateddeploymentnotes,deploymentstagerunid
Authorization: Bearer {HOST_TOKEN}

Store aigenerateddeploymentnotes as AI_DEPLOY_NOTES.

Phase 5 — Configure Deployment Settings

5.0a Check for deployment-settings.json:

Check if deployment-settings.json exists in the project root:

• If it exists: Read it and show a summary of configured stages and env var counts. Say: "Found existing deployment-settings.json with {N} stages configured." (Count top-level keys as stage names; count EnvironmentVariables entries per stage.)

• If it does NOT exist AND there are env var definitions in the solution manifest (from solutionManifest.envVars[] if available, or from the query in 5.1): Generate a template file and inform the user. Template structure:

  {
    "{stageName}": {
      "EnvironmentVariables": [
        { "SchemaName": "{envVarSchemaName}", "Value": "" }
      ],
      "ConnectionReferences": []
    }
  }
  

  Use the stage name from SELECTED_STAGE.name and env var schema names from .solution-manifest.json (if available) or from the 5.1 query. Write to deployment-settings.json at the project root. Say: "Generated deployment-settings.json template. Fill in values before deploying, or provide them now when prompted."

  Note: If env vars are not yet known at this point (5.0a runs before 5.1), generate the template file after 5.1 completes and the env vars are discovered — then inform the user before continuing to the prompt in 5.1.

• If it does NOT exist AND there are no env vars in the solution: Note "No env var overrides needed" and skip.

5.0b Surface the file path:

Always display the resolved path {projectRoot}/deployment-settings.json so the user knows where to find it, whether it was just created or already existed.

5.1 Discover env var definitions in the solution and resolve per-stage values:

Query the solution components in the source environment to find all env var definitions (componenttype 380):

GET {sourceEnvUrl}/api/data/v9.2/solutioncomponents?$filter=_solutionid_value eq '{solutionId}' and componenttype eq 380&$select=objectid
Authorization: Bearer {SOURCE_TOKEN}

For each objectid, fetch the schema name:

GET {sourceEnvUrl}/api/data/v9.2/environmentvariabledefinitions({objectid})?$select=schemaname,displayname,type,defaultvalue

This gives you SOLUTION_ENV_VARS — the list of env vars that will travel to the target.

Read deployment-settings.json (if it exists in the project root) and look up the selected stage name to get pre-configured values:

const stageSettings = deploymentSettings?.stages?.[selectedStageName] || {};
const preconfigured = stageSettings.EnvironmentVariables || []; // [{ SchemaName, Value }]

Identify unconfigured env vars — those in SOLUTION_ENV_VARS that have no entry in preconfigured:

const unconfigured = SOLUTION_ENV_VARS.filter(v =>
  !preconfigured.find(p => p.SchemaName === v.schemaname)
);

🚦 Gate (plan · deploy-pipeline:5.env-vars): Unconfigured env vars per stage — user supplies values or skips (uses default). Without values, runtime reads default which may be dev-only. In MULTI_RUN_MODE / MULTI_PIPELINE_MODE this gate fires per loop iteration that has unconfigured env vars — values supplied for iteration 1 do NOT carry to iteration 2.

If there are unconfigured env vars, present them to the user via AskUserQuestion:

"This solution has {N} environment variable(s) with no value configured for {stageName}. Enter the value for each (leave blank to use the default, or skip if not applicable):

1. {schemaname} ({displayname}) — default: {defaultvalue ?? 'none'}
2. ..."

Collect responses and merge with preconfigured to form the final ENV_VAR_OVERRIDES array. Offer to save the values back to deployment-settings.json for future runs:

"Save these values to deployment-settings.json for future deployments to {stageName}?

1. Yes — save for next time
2. No — use once only"

If Yes: write/update deployment-settings.json with the collected values under stages.{stageName}.EnvironmentVariables.

If all env vars are pre-configured (or there are none): skip the prompt, use preconfigured directly.

5.1b Pre-PATCH validation of deployment-settings.json Secret references.

Why this gate exists. The Power Platform Pipelines handler validates the deploymentsettingsjson PATCH at import time, AFTER the stage run has been queued and potentially after a long wait behind serialized imports. A bad Secret reference value — placeholder syntax like @KeyVault(vaultName=...;secretName=...), raw secret values committed to the file, malformed URIs — fails the import with "ImportAsHolding failed: The value provided as a secret reference does not match a valid secret reference format." Live evidence: a 4h41m queue wait + fail in a recent session. Catching the bad reference here turns hours of wasted compute into a sub-second hard stop with a precise remediation pointer.

Skip this step entirely when ENV_VAR_OVERRIDES is empty AND deployment-settings.json is absent — there's nothing to validate.

Otherwise, run the shared helper:

node "${PLUGIN_ROOT}/scripts/lib/validate-deployment-settings.js" \
  --settingsFile "./deployment-settings.json" \
  --envUrl "{sourceEnvUrl}" \
  --stageLabel "{SELECTED_STAGE.name}"

Capture stdout as JSON. The helper classifies each EnvironmentVariables[] entry by valueFormat (kv-uri / kv-resource-id / kv-placeholder / empty / plain-text / invalid-uri) and status (valid / invalid / unknown-type / skipped). When --envUrl is provided (as above), Secret-type entries are validated against the canonical Key Vault reference formats; other types are structurally validated.

Branch on summary:

• summary.invalid === 0 → all entries valid, proceed to Phase 5.2.

• summary.invalid > 0 → STOP. Display the invalid findings[] to the user with each entry's schemaName, valueFormat, value, and message. Then surface a remediation message:

  "Pre-deploy validation found {summary.invalid} invalid env var value(s) in deployment-settings.json. The Power Platform Pipelines handler would reject these during import (potentially after a long queue wait). Fix the file before re-running this skill.

  Canonical formats for Secret env vars:

  1. Key Vault Secret Identifier URI: https://<vault>.vault.azure.net/secrets/<name>[/<version>]
  2. Azure resource ID: /subscriptions/<sub>/resourceGroups/<rg>/providers/Microsoft.KeyVault/vaults/<vault>/secrets/<name>
  3. Empty (use the env var definition's default)

  See configure-env-variables Phase 3.B and add-server-logic Phase 7.2a for the end-to-end flow (vault selection → secret storage → URI handoff)."

  Do NOT prompt for "proceed anyway" — there's no partial-validity case where forcing the bad PATCH leads to a successful import. The deploy will fail; only the wait time changes.

• summary.invalid === 0 && summary.['unknown-type'] > 0 → log a one-line warning about the schemas whose types couldn't be looked up (probably auth or transient errors) and proceed. The downstream handler will still reject obvious garbage; the pre-PATCH gate is a fast-path, not the only line of defense.

5.2 PATCH stage run with artifact version, deployment notes, and environment variables (always run):

First, determine the current solution version in the source (dev) environment — this must match exactly:

GET {sourceEnvUrl}/api/data/v9.0/solutions?$filter=uniquename eq '{SOLUTION_NAME}'&$select=version
Authorization: Bearer {SOURCE_TOKEN}

Use the returned version as artifactdevcurrentversion. Do NOT use the version from .solution-manifest.json — that may be stale.

For artifactversion, increment the patch number of the source version (e.g., 1.0.0.2 → 1.0.0.3). This must be strictly greater than the version already deployed in the target stage. If deploying to the same stage multiple times, check docs/alm/last-deploy.json for the last artifactVersion and use a higher value.

Then PATCH (include deploymentsettingsjson only if ENV_VAR_OVERRIDES is non-empty):

PATCH {hostEnvUrl}/api/data/v9.0/deploymentstageruns({STAGE_RUN_ID})
Content-Type: application/json
Authorization: Bearer {HOST_TOKEN}

{
  "artifactdevcurrentversion": "{current version from source env — must match exactly}",
  "artifactversion": "{new version — must be > current version in target stage}",
  "deploymentnotes": "{AI_DEPLOY_NOTES if available, otherwise a brief description of what is being deployed}",
  "deploymentsettingsjson": "{JSON.stringify({ EnvironmentVariables: ENV_VAR_OVERRIDES, ConnectionReferences: [] })}"
}

The deploymentsettingsjson value must be a JSON-encoded string (double-serialized):

const deploymentsettingsjson = JSON.stringify({
  EnvironmentVariables: ENV_VAR_OVERRIDES,
  ConnectionReferences: stageSettings.ConnectionReferences || [],
});

If ENV_VAR_OVERRIDES is empty and there are no connection references, omit deploymentsettingsjson entirely.

Response is HTTP 204. If the PATCH fails with a version conflict error, check both version values and retry.

Caveat — deploymentsettingsjson does NOT always write environmentvariablevalues records on the target post-deploy. Observed live: a Power Pages site with env var definitions that are not yet linked to an mspp_sitesetting record can complete a successful deploy with a populated deploymentsettingsjson, and the target environment's environmentvariablevalues table still ends up empty for those entries. The Power Platform Pipelines handler appears to write values only for definitions that have an existing site-setting binding (or another consumer the platform recognizes). This may be a platform bug or by-design pending a separate Power Pages Management UI step. If you ship Secret env vars or any other type that the user expects to land in the target before setup-auth/configure-env-variables runs there, verify in Phase 7 below that environmentvariablevalues records exist on the target after the deploy completes, and surface a clear prompt if they don't. Workaround: invoke configure-env-variables (or run link-site-setting-to-env-var.js per definition) on the target after the deploy to create the values explicitly. Track this in the next PR if it becomes a recurring blocker.

Phase 6 — Deploy and Monitor

🚦 Gate (final · deploy-pipeline:6.0.final-consent): Last-call before each DeployPackageAsync / pac pipeline deploy call. Validation already passed. Non-transactional import — partial failure leaves whatever already imported on the target.

⚠ Fires PER LOOP ITERATION in MULTI_RUN_MODE / MULTI_PIPELINE_MODE. Three solutions in deploymentOrder → three Phase 6.0 prompts (one before each iteration's DeployPackageAsync). The upstream Phase 2 stage selection — whether via interactive AskUserQuestion or via the --stage argument — covers only stage choice, NOT individual deploy authorization. Never batch the deploy loop. Skipping the per-iteration prompt and proceeding straight from Phase 5 → Phase 6.1 for solutions 2..N is the documented strategy violation this gate exists to prevent.

6.0 Final deploy consent gate. Before kicking off each deployment — whether via DeployPackageAsync (6.1) or the pac pipeline deploy PAC CLI fallback, and whether this is the first solution or the Nth in MULTI_RUN_MODE — confirm with the user explicitly. The earlier gates (Phase 2 stage pick, Phase 2.5 unblock, Phase 3.5 completeness, Phase 5 env var values) each cover their own decision, but none of them is a per-iteration "ready to ship this one?" prompt and the deploy itself is not transactional — partial failures leave whatever has already imported on the target. This gate makes every production-promotion moment explicit. Use AskUserQuestion:

"Ready to deploy {ARTIFACT_SOLUTION_NAME} (v{newVersion}) to {SELECTED_STAGE.name} ({targetEnvUrl})?

This will run for ~3–60 minutes depending on solution size. The import is not transactional — if it fails partway, whatever already imported stays on the target and a manual cleanup (or re-deploy with a higher artifact version) is required to recover. docs/alm/last-deploy.json is written regardless of outcome.

1. Deploy now — POST DeployPackageAsync (or run pac pipeline deploy for the CLI fallback) and poll until terminal
2. Cancel — leave the stage run in its current state (validated but not deployed); re-run /power-pages:deploy-pipeline later when ready"

Branch on the answer:

• Deploy now → proceed to 6.1 below (or the PAC CLI fallback box).
• Cancel → stop cleanly. Do NOT call DeployPackageAsync or pac pipeline deploy. Tell the user: "Cancelled. The stage run {STAGE_RUN_ID} is validated but not deployed. Re-invoke /power-pages:deploy-pipeline to resume. If the source solution version changes before you re-invoke, the stage run's artifactversion field will need a fresh PATCH (Phase 5.2) before deploying — re-running the skill handles that automatically." Do not write a docs/alm/last-deploy.json on cancel — there's no deploy outcome to record.

For Production targets specifically, treat this gate as the single most important prompt in the skill — the cancellation cost (stage run already validated) is small compared to a wrong-stage import.

If ValidatePackageAsync was unavailable (VALIDATE_PACKAGE_UNAVAILABLE = true): use the PAC CLI as the primary deployment mechanism instead of 6.1. The same Phase 6.0 consent gate above gates this path too — do not run pac pipeline deploy until the user has answered "Deploy now":

Ask user for currentVersion (pre-fill from .solution-manifest.json solution.version if available) and newVersion (suggest incrementing the patch number, e.g. 1.0.0.0 → 1.0.0.1).

pac pipeline deploy \
  --environment "{devEnvUrl}" \
  --solutionName "{ARTIFACT_SOLUTION_NAME}" \
  --stageId "{SELECTED_STAGE.stageId}" \
  --currentVersion "{currentSolutionVersion}" \
  --newVersion "{newVersion}" \
  --wait

If the CLI returns "Resource not found for the segment 'deploymentenvironments'": the dev environment is not connected to a Pipelines host. Advise the user to configure the host in Power Platform Admin Center (Environments → select env → Pipelines), then retry.

If CLI succeeds: parse the output for stage run status, write docs/alm/last-deploy.json with status: "Succeeded" (or the parsed status), and skip the DeployPackageAsync call and polling in 6.1–6.2.

6.1 Trigger deployment (skip if VALIDATE_PACKAGE_UNAVAILABLE = true — use PAC CLI path above). Only run after the Phase 6.0 consent gate returned "Deploy now":

POST {hostEnvUrl}/api/data/v9.0/DeployPackageAsync
Content-Type: application/json
Authorization: Bearer {HOST_TOKEN}

{"StageRunId": "{STAGE_RUN_ID}"}

Note: DeployPackageAsync also returns 404 on older Pipelines package versions. If this occurs, use the pac pipeline deploy CLI path above.

6.2 Poll stagerunstatus until terminal using poll-deployment-status.js:

node "${PLUGIN_ROOT}/scripts/lib/poll-deployment-status.js" \
  --hostEnvUrl "{hostEnvUrl}" \
  --token "{HOST_TOKEN}" \
  --stageRunId "{STAGE_RUN_ID}" \
  --intervalMs 8000 \
  --maxAttempts 75

Capture stdout as JSON: const result = JSON.parse(output).

The script polls msdyn_stagerunstatus until a terminal state:

• result.status === 'Succeeded' → proceed to Phase 7
• result.status === 'Awaiting' → approval gate (see below); do NOT treat as error
• Script throws on 200000003 (Failed) or 200000004 (Canceled) — error message includes msdyn_errordetails
• Script throws on timeout after ~10 minutes

suboperation field (not polled by the script but visible in Power Platform) shows progress detail:

• 200000100 = None (starting/finishing)
• 200000105 = Deploying Artifact (actively installing solution)

🚦 Gate (pause · deploy-pipeline:6.pending-approval): External wait — PP Pipelines stagerunstatus=200000005 mid-deploy. User approves in PPAC. Cancel here PATCHes iscanceled: true on the run. In MULTI_RUN_MODE / MULTI_PIPELINE_MODE this gate fires per loop iteration that hits Pending Approval.

Approval gate handling: If result.status === 'Awaiting' (msdyn_stagerunstatus = 200000005):

• Inform user: "This deployment is waiting for approval. Please approve it in Power Platform: make.powerapps.com → Solutions → Pipelines → find deployment for {STAGE_RUN_ID} → Approve."
• Ask via AskUserQuestion: "Have you approved the deployment? 1. Yes, I approved it — continue polling / 2. Cancel deployment"
• If Yes: re-run poll-deployment-status.js to continue polling.
• If Cancel: PATCH the stage run to cancel it:

  PATCH {hostEnvUrl}/api/data/v9.0/deploymentstageruns({STAGE_RUN_ID})
  {"iscanceled": true}
  

  Then record status as "Canceled".

Token refresh: After every 10 poll cycles (~80 seconds), refresh HOST_TOKEN via az account get-access-token and pass the updated token in a fresh poll-deployment-status.js invocation with reduced --maxAttempts.

Report deployment progress updates as polling continues.

Phase 7 — Write Deployment Record and Summary

7.1 Determine final status string:

• 200000002 → "Succeeded"
• 200000003 → "Failed"
• 200000004 → "Canceled"
• 200000005 → "PendingApproval" (if user cancelled waiting)
• Poll timeout → "Unknown"

7.2 Post-deployment warnings (only if deployment Succeeded):

Using solutionManifest captured in Phase 1 from detect-project-context.js, check for components that require manual follow-up in the target environment.

Connection reference warning — if solutionManifest.cloudFlows is present and non-empty:

⚠️ Connection references may need binding This solution includes cloud flow(s). If those flows use connection references (e.g. Dataverse, SharePoint), they must be bound to live connections in the target environment or the flows will remain disabled.

To bind: Power Automate → target environment → each flow → Edit → bind connections.

If solutionManifest.cloudFlows is absent or empty, skip this warning entirely.

Bot republish warning — if solutionManifest.botComponents is present and non-empty:

⚠️ Bot republish required This solution includes a Copilot Studio bot. After deployment, the bot must be republished in the target environment to complete provisioning.

To republish: Power Pages Management → target environment → Edit site → Copilot → republish.

If solutionManifest.botComponents is absent or empty, skip this warning entirely.

These warnings are informational only — do not block the summary or use AskUserQuestion.

7.3 Write docs/alm/last-deploy.json (create the docs/alm/ directory first if missing — node -e "require('fs').mkdirSync('docs/alm',{recursive:true})"):

{
  "pipelineId": "{pipelineId}",
  "pipelineName": "{pipelineName}",
  "stageId": "{SELECTED_STAGE.stageId}",
  "stageRunId": "{STAGE_RUN_ID}",
  "stageName": "{SELECTED_STAGE.name}",
  "solutionName": "{ARTIFACT_SOLUTION_NAME}",
  "solutionId": "{ARTIFACT_SOLUTION_ID}",
  "status": "{final status string}",
  "deployedAt": "{ISO timestamp}",
  "hostEnvUrl": "{hostEnvUrl}",
  "targetEnvironmentUrl": "{SELECTED_STAGE.targetEnvironmentUrl}",
  "artifactVersion": "{artifactVersion from Phase 5.2 PATCH}",
  "deployHistoryFile": "docs/deploy-history/{YYYY-MM-DD}-{stageName}-{artifactVersion}.html",
  "activationStatus": null,
  "siteUrl": null
}

activationStatus and siteUrl start as null and are patched at the end of Phase 7.7 once the activation outcome is known.

Where {YYYY-MM-DD} is the date portion of deployedAt and {stageName} is the stage name with spaces replaced by hyphens (lowercased), e.g. 2026-04-06-staging-1.0.0.3.md.

Retry attempts go as PEER entries in runs[], not nested under runs[i].lastAttempt. In MULTI_RUN_MODE (multi-solution) the marker carries a runs[] array — one entry per solution per attempt. When the user retries a failed solution (e.g. after fixing an env var value, re-running deploy-pipeline against the same stage), the retry is a NEW peer entry in runs[] with its own artifactVersion and attemptedAt. Do NOT nest the retry as runs[i].lastAttempt: {...} under the original failed entry — that mixes the schema and makes audit traversal ambiguous (was 1.0.0.4 the deployed version or just the latest attempt?). The flat peer-array shape lets every consumer (refresh-alm-plan-data, the rendered Pipelines tab, the deploy history HTML) walk runs[] once and see the full timeline.

Marker writers in this skill should append, not mutate. Each retry of a failed solution writes a new runs[] entry with status='Succeeded' (or 'Failed' on persistent failure) and the post-bump artifactVersion. The original failed entry stays untouched as history.

7.4 Write deployment history entry (HTML):

Compute the history filename: {YYYY-MM-DD}-{stageName}-{artifactVersion}.html (same derivation as docs/alm/last-deploy.json's deployHistoryFile field — replace spaces with hyphens, lowercase stage name).

Create docs/deploy-history/ if it does not already exist:

mkdir -p docs/deploy-history

Read the template at ${PLUGIN_ROOT}/skills/deploy-pipeline/assets/deploy-history-template.html and replace the following __PLACEHOLDER__ tokens:

Overview tab:

Placeholder	Value
__SOLUTION_FRIENDLY_NAME__	Solution friendly name (from .solution-manifest.json) or {solutionUniqueName}
__SOLUTION_NAME__	{ARTIFACT_SOLUTION_NAME}
__STAGE_NAME__	{SELECTED_STAGE.name}
__TARGET_ENV_URL__	{SELECTED_STAGE.targetEnvironmentUrl}
__STAGE_RUN_ID__	{STAGE_RUN_ID}
__PIPELINE_NAME__	{pipelineName}
__DEPLOYED_AT__	{deployedAt ISO string}
__ARTIFACT_VERSION__	{artifactVersion from Phase 5.2}
__PREV_ARTIFACT_VERSION__	{artifactDevCurrentVersion} — the version that was in dev before this deploy
__STATUS_CLASS__	succeeded / failed / pending-approval
__STATUS_ICON__	✓ for Succeeded, ✗ for Failed, ⏳ for PendingApproval
__STATUS_LABEL__	Succeeded / Failed / Pending Approval
__ACTIVATION_SECTION__	Initially '' — replaced in Phase 7.7 once activation outcome is known

Solution tab — read .solution-manifest.json to build these sections:

Placeholder	Value
__SOLUTION_META_ROWS__	<tr> rows for: Friendly Name, Unique Name, Version (new → previous), Type (Managed/Unmanaged), Publisher, Total Components. Source: manifest + validationResults.SolutionDetails from Phase 6.
__VALIDATION_SECTION__	If validation passed: <div class="note-box succeeded"><span class="validation-badge passed">✓ Validation Passed</span> — No missing dependencies.</div>. If failed or deps present: <div class="note-box warning"> listing each missing dependency name.
__SOLUTION_CONTENTS_SECTION__	Build from .solution-manifest.json: a <div class="contents-grid"> with two <div class="contents-card"> blocks — Dataverse Tables (as <span class="table-chip"> per table) and Bot Components (comma-separated names). Below the grid, add a <div class="note-box neutral"> with: {totalAdded} components added to solution (from components.totalAdded). If manifest is unavailable, show a neutral note.

Config & Notes tab:

Placeholder	Value
__ENV_VARS_SECTION__	If ENV_VAR_OVERRIDES was non-empty: a <div class="card"><h3>Environment Variable Overrides</h3> table with schema name + override value columns. Otherwise: <div class="note-box neutral">No environment variable overrides applied.</div>
__DEPLOYMENT_NOTES_SECTION__	If AI_DEPLOY_NOTES is available: <div class="card"><h3>AI Deployment Notes</h3><p>…</p></div>. Otherwise: ''
__POST_DEPLOY_WARNINGS__	One <div class="note-box warning"> per post-deploy warning (connection refs, bot republish). Empty string if none.

Write the rendered HTML to docs/deploy-history/{filename}.html.

Then add to the staging area:

git add docs/alm/last-deploy.json docs/deploy-history/{filename}.html
git commit -m "Deploy {solutionUniqueName} v{artifactVersion} to {stageName} ({status})"

If git is not initialized in the project root (i.e., git rev-parse --git-dir fails), skip the commit silently.

7.5 Record skill usage:

Reference: ${PLUGIN_ROOT}/references/skill-tracking-reference.md

Follow the skill tracking instructions in the reference to record this skill's usage. Use --skillName "DeployPipeline".

7.5b Refresh the ALM plan (if one exists):

node "${PLUGIN_ROOT}/scripts/lib/refresh-alm-plan-data.js" \
  --projectRoot "." \
  --phase deploy-pipeline \
  --render

The helper reads the docs/alm/last-deploy.json you just wrote, ingests it into planData.pipelineMeta.lastDeploy, drops any pre-deploy "host not yet provisioned" risks, and re-renders docs/alm-plan.html so the Pipelines tab shows the actual run state (status, version, component count, activation, site URL). When docs/.alm-plan-data.json is absent (the skill was invoked standalone, not part of an ALM plan), the helper returns ok:false as a soft no-op — safe to run unconditionally.

This step is what keeps the rendered plan current — plan-alm is a planner and does not refresh the plan itself, so each execution skill owns its own post-run refresh. Running it more than once is idempotent (same input → same output).

Point the user at the next step (user-driven sequencing). The helper's stdout JSON includes nextStep: { name, skill: string | null } | null. When non-null, branch on skill: when skill is non-null, tell the user "Plan updated. Next in your plan: {nextStep.name} → run {nextStep.skill} when you're ready."; when skill is null (an internal step such as Finalize, no user command), name the step only — "Plan updated. Next in your plan: {nextStep.name}." — and never print run null. (For a multi-stage pipeline this is typically the next stage's deploy, or the next stage's activate/test if those are separate steps.) When null (all steps done) or the helper returned ok:false (no plan), say nothing about a next step. Never auto-invoke the next skill — the user drives execution.

7.6 Present summary:

If Succeeded:

✓ Deployment succeeded

  Solution:     {solutionName}
  Stage:        {stageName}
  Target:       {targetEnvironmentUrl}
  Completed at: {deployedAt}
  Stage run ID: {STAGE_RUN_ID}
  Site URL:     {siteUrl from 7.7, or "— activation pending" if not yet activated, or "— checking…" before 7.7 runs}

If Failed:

✗ Deployment failed

  Stage run ID: {STAGE_RUN_ID}
  Status:       Failed

  To investigate: open Power Platform make.powerapps.com → Solutions → Pipelines
  and find the failed run for details on what caused the failure.

7.6.1 Diagnose the failure. Before asking the user what to do, query the stage run record for error details so we can offer a targeted remediation prompt:

GET {hostEnvUrl}/api/data/v9.1/deploymentstageruns({STAGE_RUN_ID})?$select=errordetails,validationresults,stagerunstatus
Authorization: Bearer {HOST_TOKEN}

Parse errordetails and validationresults as JSON / text. Check for these patterns (case-insensitive):

Pattern in error text	Diagnosis	Remediation prompt
AttachmentBlocked OR -2147188706 OR not a valid type OR \.js.*blocked	Blocked attachments — the target env's blockedattachments setting rejects file types in the solution	See 7.6.2 below
secret reference does not match a valid OR ImportAsHolding failed.*secret reference OR KeyVault.*format	Invalid Secret reference — deployment-settings.json carries a Secret value in a non-canonical format (e.g. @KeyVault(vaultName=...;secretName=...), raw secret text, malformed URI)	See 7.6.4 below
MissingDependency OR Missing dependency	Missing solution dependencies in target	Surface the dependencies; recommend the user install them and retry
(anything else)	Unknown failure	Fall through to the generic retry/exit prompt

7.6.2 Blocked-attachment remediation (gated). Only run when the diagnostic in 7.6.1 matched the blocked-attachment pattern. Never modify the env's blockedattachments setting without an explicit AskUserQuestion gate. This is a tenant-wide security setting; auto-modifying it would silently weaken security posture across other apps in the same env.

1. Switch PAC CLI to the target environment so the helper queries the right env's settings, then identify which file types are blocked. By default the helper checks js; pass --extensions if the error mentions other types (e.g. js,css):

   pac env select --environment "{TARGET_ENV_URL}"
   node "${PLUGIN_ROOT}/scripts/lib/fix-blocked-attachments.js" \
     --envUrl "{TARGET_ENV_URL}" \
     --extensions js \
     --dry-run
   

   Read the output JSON. The fields you care about: wasBlocked[] (the extensions currently blocked on the env that were on your --extensions list — these are what would be removed) and unchanged[] (extensions you asked to remove that aren't actually blocked, no-op).