Assemble Ppmplugin logo

Assemble Ppmplugin

Organization
microsoft
assemble-ppmplugin

Assemble the final `.ppmplugin` binary bundle for a PAM native extension and verify its contents. First **reconciles** the manifest's declared `entrypoints` against the binaries actually staged — if the manifest declares a platform with no built binary it gates (build it / ship without it / stop) rather than shipping a broken bundle. Then re-runs the plugin's upload-compatibility checks on the reconciled manifest, zips the manifest plus whichever of `android/<PascalName>Plugin.dex` and `ios/<PascalName>Plugin.framework/` are present into `ppmplugin/<name>.ppmplugin`, and verifies the archive layout with `jar tf` (exactly the manifest + the shipped binaries — nothing missing, nothing extra). Output: a statically verified `.ppmplugin` file on disk. Prereqs: `jar` (JDK). Run after /generate-ppmplugin-manifest and the build skill(s).

Overview

Publishermicrosoft
Repositorypower-platform-skills
Skill nameassemble-ppmplugin
Stars
895
Forks
182
Bundled files
Instructions only
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by microsoft on GitHub. Read the source before you install it.

Installation

Install the Assemble Ppmplugin AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/microsoft/power-platform-skills.git /tmp/power-platform-skills
mkdir -p .claude/skills
cp -r /tmp/power-platform-skills/plugins/power-apps-mobile-extension/skills/assemble-ppmplugin .claude/skills/assemble-ppmplugin
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Assemble Ppmplugin in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Assemble Ppmplugin on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Assemble Ppmplugin is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

/assemble-ppmplugin

The final step: take the label (manifest.json from /generate-ppmplugin-manifest) and the binaries the build skills produced, zip them into the single <name>.ppmplugin file, and verify the box contains exactly the right items. The output is the deliverable — the file you can hand off / upload. Before zipping, it reconciles the manifest's declared platforms against the binaries that actually built, so the shipped bundle never claims a platform it doesn't contain.

Read shared/ppmplugin-format.md §1 — the bundle layout this skill produces and verifies.

What this skill does NOT do

  • Does not author the manifest or build any binary — it consumes the staged outputs of the prior third-party-control skills (manifest + the build skill(s)).
  • Does not upload to Dataverse / wire into a canvas app (Stage 3 — deferred).
  • Does not build any binary — it consumes whatever the build skills staged (android/<Pascal>Plugin.dex from /build-android-binary and/or ios/<Pascal>Plugin.framework/ from /build-ios-binary) and bundles those present.

Step 1 — Read shared docs + prereq block

  1. Read shared/shared-instructions.md and shared/ppmplugin-format.md.
  2. Prereq: jar (ships with the JDK). Print the visible block (shared-instructions §9.2):
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 Prereq check — /assemble-ppmplugin
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 🟢 ✓ jar (JDK) available
 🟢 1 check passed, 0 failed. Ready to proceed.

If jar is missing, STOP with BLOCKED: jar not found — install a JDK (JDK 17).


Step 2 — Reconcile declared entrypoints against staged binaries (the gate)

The shipped bundle's manifest.json MUST declare exactly the platforms whose binaries are in the zip — no more, no less. The manifest was authored from intent (target choice); this step makes it match reality (what actually built). This is the one place that sees the final set of binaries, so it owns consistency.

  1. Read ppmplugin/staging/manifest.json — else STOP with NEEDS_CONTEXT: manifest missing — run /generate-ppmplugin-manifest. Read name for the output filename <name>.ppmplugin (the version is NOT in the filename — it lives in the manifest's version field; the wrap pipeline reads it from there. Still read version to show it in the deliverable report).
  2. Declared platforms: is entrypoints.android present? entrypoints.ios?
  3. Staged binaries:
    • Android: ppmplugin/staging/android/<dex> exists, where <dex> = entrypoints.android.dex.
    • iOS: ppmplugin/staging/ios/<framework>.framework/ exists (a flat .framework, NOT an .xcframework§5b), where <framework> = entrypoints.ios.framework. If an .xcframework is staged instead, surface it as a gate — the wrap CI can't ingest it; re-run /build-ios-binary for the flat device-slice framework.
  4. Reconcile, per platform:
DeclaredStagedAction
yesyesship it ✓
yesnoMISMATCH → gate (below)
noyesbinary present but the manifest won't route it — surface; offer to re-run /generate-ppmplugin-manifest to declare it, or leave it out
nononot part of this bundle — ignore
  1. If nothing is both declared AND staged → STOP with NEEDS_CONTEXT: no built binary to ship — run a build skill first.

The mismatch gate (declared but not staged). Do NOT silently ship a manifest pointing at a missing binary. The situation is genuinely ambiguous (changed-my-mind vs forgot-to-build), so surface it and ask via AskUserQuestion:

"Manifest declares <platform> but no <platform> binary is staged."

  • Build <platform> first — run the build skill (/build-android-binary, or /build-ios-binary for iOS, Mac-only) [invoke it via the Skill tool — execute, don't describe]
  • Ship without <platform> — remove entrypoints.<platform> from the staged manifest.json, ship the platforms that ARE built
  • Stop

If the user picks "Ship without", edit the staged manifest.json to drop that entrypoint before continuing. After this step, the manifest's entrypoints set equals the staged-binary set exactly.


Step 3 — Final validation pass

Re-run the ppmplugin-format §4 validation rules on the reconciled staged manifest.json (it may have been edited in Step 2 or by hand). Also cross-check, for each shipped platform:

  • Android: entrypoints.android.dex exactly matches the staged DEX filename.
  • iOS: entrypoints.ios.framework matches the staged …/<framework>.framework directory name, and that .framework/ carries the binary + Headers/<framework>.h + Modules/module.modulemap + Info.plist (the wrap-CI requirements — /audit-ppmplugin does the deep check; flag here if any are obviously absent).

If validation fails, STOP with BLOCKED: manifest validation — <rule> and do not produce the bundle. A .ppmplugin that fails the local compatibility checks is not ready for upload.


Step 4 — Zip + verify

Before zipping: if the output ppmplugin/<name>.ppmplugin already exists, do NOT silently overwrite it — ask via AskUserQuestion: Replace [default] / Keep existing (stop, leave the prior bundle untouched). Proceed to zip only on Replace.

  1. Zip the manifest plus only the platform folder(s) reconciled as shipped in Step 2. Pass them explicitly so the build dirs (android-build/, ios-build/) are never swept in. jar (JDK) recurses directories, so it handles the iOS .framework tree, and it behaves identically on every OS — only the shell glue differs (§5):

    macOS / Linux (bash):

    bash
    mkdir -p ppmplugin
    # list only the shipped platforms, e.g. `android` for Android-only, `android ios` for Both:
    ( cd ppmplugin/staging && jar cMf "../<name>.ppmplugin" manifest.json android ios )

    Windows (PowerShell):

    powershell
    New-Item -ItemType Directory -Force ppmplugin | Out-Null
    Push-Location ppmplugin\staging
    jar cMf ..\<name>.ppmplugin manifest.json android ios
    Pop-Location

    (-M = no JAR manifest entry; the bundle's manifest.json is our own. Drop ios (or android) from the argument list if that platform isn't shipped.) On Replace (confirmed above), the prior bundle is overwritten.

  2. Verify layout — list and assert it matches the reconciled manifest exactly:

    bash
    jar tf ppmplugin/<name>.ppmplugin

    Android-only expects exactly:

    manifest.json
    android/<Pascal>Plugin.dex

    A Both bundle also has ios/<Pascal>Plugin.framework/… entries (flat framework — binary + Headers/ + Modules/module.modulemap + Info.plist). STOP with BLOCKED: unexpected bundle contents (and the actual listing) if: a build dir leaked in (android-build/, ios-build/), a stray file appears (.DS_Store, META-INF/), an ios/<…>.xcframework/ is present (must be a flat .framework), or a declared platform's binary is missing. The bundle must contain exactly the manifest + the entrypoints' binaries — nothing else.


Step 5 — Report

Update the ## ppmplugin (third-party controls) block in .extension-state.md: Bundle: <name>.ppmplugin assembled <ISO timestamp> (v<version>, platforms: <android[,ios]>). Print the deliverable as a visible fenced block (shared-instructions §9.1, informational form):

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 .ppmplugin ready
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 ppmplugin/pen-input.ppmplugin   (v0.1.4, Android-only)
 Contents:
   manifest.json
   android/PenInputPlugin.dex
 Next: /audit-ppmplugin — verify the bundle is upload-ready before shipping it.
       (Then: deploy the dispatcher PCF via /publish-pcf-companion, and upload the
        .ppmplugin via the wrap wizard — Stage 3, not yet a skill.)

Then offer the next step via AskUserQuestion (shared-instructions §9.1 — invoke the chosen skill via the Skill tool; execute, don't describe):

  • Run /audit-ppmplugin (recommended — the final upload-readiness gate: validator rules + DEX SDK-leakage scan + iOS framework checks)
  • Run /publish-pcf-companion (deploy the dispatcher PCF to a Power Platform env)
  • Stay — I'll inspect the bundle first

(When /assemble-ppmplugin runs as an internal stage of /generate-ppmplugin, the orchestrator invokes /audit-ppmplugin automatically and this gate is skipped.) Return DONE with the bundle path.

Frequently asked questions

What does the Assemble Ppmplugin AI skill do?

Assemble the final `.ppmplugin` binary bundle for a PAM native extension and verify its contents. First **reconciles** the manifest's declared `entrypoints` against the binaries actually staged — if the manifest declares a platform with no built binary it gates (build it / ship without it / stop) rather than shipping a broken bundle. Then re-runs the plugin's upload-compatibility checks on the reconciled manifest, zips the manifest plus whichever of `android/<PascalName>Plugin.dex` and `ios/<PascalName>Plugin.framework/` are present into `ppmplugin/<name>.ppmplugin`, and verifies the archiv...

Why use Assemble Ppmplugin on TypingMind?

Because you install it once and use it with any model. Assemble Ppmplugin is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Assemble Ppmplugin in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/microsoft/power-platform-skills/tree/main/plugins/power-apps-mobile-extension/skills/assemble-ppmplugin. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Assemble Ppmplugin?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Assemble Ppmplugin?

As many as you like. As long as a model supports skills, you can use Assemble Ppmplugin with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Assemble Ppmplugin AI skill free?

Yes. It is published on GitHub by microsoft under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇