Loading…
Loading…
power-platform · an Agent Skill by SkillEra for Claude Code, Codex, Cursor and any agent that reads SKILL.md
Power Platform development the way software is built, and proved in the product. Canvas apps, Dataverse solutions and Power Automate flows with the definition in git, a portable artifact built from it, a deliberate deployment, and every change proved by performing the task in the published app, driven by Playwright.
With guidance for Power Pages sites and for Power BI as an app's reporting layer.
This skill is a working method built from those failures. Every trap it describes comes with its signature, so a symptom is recognised in minutes instead of days, and every step of shipping states what it proves and what it does not.
It loads on its own for Power Platform tasks — a Save button that works in Studio but does nothing in the published app, a picker over a large Dataverse table, solution flow JSON, or setting up a repository for canvas app and Dataverse work.
Described from the README at v0.18.0 (opens in a new tab) — a newer tagged version is available on GitHub, and may change what follows.
SKILL.md routes each task to one reference. Each topic has one home; where another reference touches it, it points there rather than repeating it.
check-canvas-overlap — A control placed over another that can be on screen at the same time, or decoration declared over a button or label.
Power Pages coverage comes from one site.
deploy-tables — Creates the publisher, solution, tables, columns, choice options and lookups described in a JSON manifest, then reads every table, column, option and lookup back from the environment. It only adds: a re-run is a no-op, nothing is renamed, retyped or deleted.
Every non-trivial change moves through the same cycle. Each step proves one thing, and skipping one is how a change reaches users unproven.
The method and references need nothing installed. The tools need Node 20 or later, Python 3 and the Power Platform CLI, and the canvas authoring server needs the .NET 10 SDK. Playwright is optional, for browser verification; PowerShell is optional, for the token and owner scripts.
Every checker carries a --selftest that proves it goes red on a known-bad input, and CI runs them on every push.
Measured on version 0.7.0 with three runs per configuration. Across the ten tasks, the same model passed:
The same model and prompt ran with the skill and without it, in fresh sessions, graded against checks fixed before the runs. The three canvas tasks (search and picker, long text, list filters) were re-run on 0.7.1.
The loop and recipient tasks were rewritten for 0.7.0 to be harder: a cycle hidden among five flows with a one-day drift, and a leak inside one branch of an existing flow.
The remaining misses with the skill are two flaky control-shape checks in the picker task and one note that does not state its trade-off. The cost is real: about 2.9 times the tokens, 1.2 times the output, and about 26 seconds more per task. Nothing ran against a live tenant.
The report linked below still shows an earlier measurement.
A canvas app has no test framework. Opening Studio, adding a data source, capturing a network trace and running the app are all browser work, and a browser can be driven. Verifying in the published app is the standard here, not a fallback.
The skill never installs anything silently. Without a browser, the change is reported as unverified, not passed.
“Saved” on screen proves nothing about the row: confirm reads it back from Dataverse after the steps, and a row that did not change during this run fails.
Then open your Power Platform repository in your agent and ask for real work.
Or copy skills/power-platform/ to ~/.claude/skills/power-platform/, or to the equivalent skills folder for your agent.
The harness is optional and recommended, and the agent offers it: from the project folder, setup-harness.mjs shows its plan, and --apply installs the hooks, the tools, the config and the continuity documents.
To update, run the same install again (for the plugin, /plugin marketplace update power-platform-skill), then re-run setup-harness.mjs in each project. Projects with the harness installed are told at session start, at most once a day, when a newer release exists.
Starting on a new machine or a new environment? first-run.md (opens in a new tab) is the order to set everything up in, with a smoke test to run before the first change.
Claude Code, Codex, Cursor and others, through the skills installer.
npx skills add KDavidP1987/power-platform-skillAdd the marketplace once, then install the plugin.
/plugin marketplace add KDavidP1987/power-platform-skill
/plugin install power-platform@power-platform-skillThe earlier evaluation report shows every check of that measurement, what went wrong without the skill, and the limits of the measurement; the current figures are in the README. The skill is MIT licensed; issues and pull requests are welcome on GitHub.