Power Platform Guardian — Development skill for Claude Code

Power Platform Guardian

Claude Skills that catch Power Platform problems before they hit production — delegation failures, performance traps, licensing surprises, silent error-handling gaps, offline sync bugs, and more.

Built from real, live research at the time of writing — not from AI training-data memory — and organized so each skill knows when its own cached knowledge might be stale and needs a live check before it's trusted.

Skills in this repo

Skill	Status	What it covers
`power-apps-guardian/`	Complete	Canvas app Power Fx — delegation, performance, licensing, error handling, column/data-type quirks, variable scope, offline/mobile sync, ALM/environments, component reuse, accessibility
`power-automate-guardian/`	Not started	Flow-specific pain points — Apply to Each concurrency, trigger thresholds/pagination, connection references in flows, throttling, retry policies

Each skill is self-contained — its own `SKILL.md`, its own `references/` folder, its own README. They share the same design philosophy (see below) but don't depend on each other.

Design philosophy

1. Research-first, not memory-first. Every reference file was built by searching current documentation and community-reported behavior, not generated from an LLM's training data. Where something is genuinely settled platform mechanics, it's cached for speed. Where it's the kind of thing Microsoft changes (pricing, quotas, plan names, connector classifications), it's flagged for live verification instead.
2. Consistent, checkable structure. Reference files use a shared schema (pattern → why it breaks → fix → scope → severity → live-search-trigger) so nothing gets applied inconsistently, and so a human reviewing the file can sanity-check it without reading prose.
3. Honest about incompleteness. A skill's SKILL.md says plainly what it does and doesn't cover. No fil