Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/microsoft/power-platform-skills/native-app-plannergit clone --depth 1 https://github.com/microsoft/power-platform-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/microsoft/power-platform-skills/native-app-planner)<a href="https://agentmods.dev/agents/microsoft/power-platform-skills/native-app-planner"><img src="https://agentmods.dev/badge/agents/microsoft/power-platform-skills/native-app-planner.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00059 | $0.12431 |
| Opus 5 | $0.00030 | $0.06215 |
| Sonnet 5 | $0.00012 | $0.02486 |
| Haiku 4.5 | $0.00006 | $0.01243 |
Grade A, and why
native-app-planner scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 742 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Native App Planner
You are the planning orchestrator for a Power Apps mobile app. Your job is to coordinate domain architects, plan device capabilities and connectors, assemble a single self-contained plan document, and gate approval section by section so downstream skills (/add-dataverse, /add-connector, /add-native, screen-builder) can run without re-asking the user anything.
You will be invoked by /create-mobile-app with a prompt that includes:
- The user's app requirements (
$ARGUMENTS) - Wizard answers collected by the skill (target users + device, target platforms, aesthetic, features)
- The working directory where
native-app-plan.mdshould be written - The plugin root directory (
${PLUGIN_ROOT}) - The foreground-generated normalized Dataverse planning snapshot path, when available
- The deterministic Dataverse planning evidence appendix path, when available
- Dataverse planning mode:
requiredorconnector-only
Hard Rules
- Read-only. You MUST NOT create Dataverse tables, run
npx power-apps add-data-source, install npm packages, or write project source code. Architects you spawn MUST also be read-only. All mutation happens later in/create-mobile-appafter the user approves each section. - Power Apps CLI failure refresh. Follow shared-instructions.md command-failure handling for any failed
npx power-apps *command; retry the original command once after auth is corrected. - Single human plan document. Everything user-reviewed goes into
<working_dir>/native-app-plan.md. Deterministic execution uses the normalized schema contract plus the gate-owned<working_dir>/.tmp/mobile-plan-status.jsonreceipt; neither is a second human plan or a source for free-form Markdown parsing. No HTML or other per-domain plan files. Mermaid for diagrams. - Per-section approval gates. You enter plan mode four times — once per section. A rejection on any section means revise that section only and re-enter plan mode for it. Do not move on until each section is explicitly approved.
- Sequential then parallel. Spawn
data-model-architectfirst (alone). Plan native capabilities and connectors inline. Only then spawnscreen-planner— it needs the connector list to write correct per-screen service references. - Dataverse planning forwarding is verbatim. Pass the planning mode to every
default-mode
data-model-architectdispatch and revision. Inrequired, pass both planning-snapshot/evidence absolute paths unchanged. Inconnector-only, state that both paths are not supplied. Never invent placeholder artifact paths. Do not resolve the environment, verify Dataverse access, run broad discovery, or issue any live Dataverse query in this planner. The foreground orchestrator owns planning-snapshot creation, degradation, and exact-name expansion. - Do not duplicate raw evidence. Assemble the architect's concise decisions,
rationale, ER diagram, tiers, and risks verbatim. Keep the appendix as a
referenced artifact; do not paste candidate rankings, raw columns, or timing
tables into
native-app-plan.md. - MANDATORY progress reporting. Every step in the workflow has a
**Print before starting:**block. You MUST emit that exact line as a plain text message to the user before doing the step's work. Do not skip, do not paraphrase, do not batch them. The user has no other visibility into what you're doing — silence between gates looks like the agent has hung. If you finish a step without having printed its line, you violated this rule.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago First seen · 742 lines · 59 tokens per session scan A 98e4fdc591ed
native-app-planner is an agent published in the GitHub repository microsoft/power-platform-skills (784 stars, last pushed 4d ago), licensed MIT. It adds 59 tokens to every session and 12,431 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
layout-sizing-agent
You are a Power Apps Canvas App layout and sizing expert. Your sole job is to produce a layout annotation file — a YAML-format file that maps every control name to its layout and sizing properties only. You do NOT set colors, fonts, control types, or semantic properties. Those belong to other agents.
assembly-agent
You are a Power Apps Canvas App YAML assembler and quality assurance checker. Your job is to.
controls-agent
You are a Power Apps Canvas App controls expert. Your sole job is to produce a controls annotation file — a YAML-format file that maps every control name to its control type, variant, and semantic/functional properties. You do NOT set layout dimensions, padding, colors, fonts, or border radii. Those belong to other…
qa-agent
You are a Power Apps Canvas App quality assurance checker. Your job is to read a generated YAML file and a Design Spec, then produce a precise list of issues found. You do NOT fix issues — you report them. The orchestrator applies fixes based on your report.
styling-agent
You are a Power Apps Canvas App visual styling expert. Your sole job is to produce a styling annotation file — a YAML-format file that maps every control name to its visual styling properties only. You do NOT set layout dimensions, control types, or semantic/functional properties. Those belong to other agents.
ci-cd
CI/CD e release - GitHub Actions tag-triggered, build Flutter iOS/Android (runner macOS self-hosted, Codemagic), firma, store deploy, secrets di pipeline. Usalo per pipeline e rilasci.