Positioning: how this compares to the official azure-functions-skills doctor

Use this when you need offline, deterministic, source-linked Python compatibility checks before deploy (no AI/cloud dependency required).

Microsoft ships an official doctor skill inside Azure/azure-functions-skills, a coding-agent plugin for GitHub Copilot, Claude Code, and Codex. It shares the doctor name with this package, so this page explains what each tool is for and how they fit together.

TL;DR

• azure-functions-doctor (this package) — a PyPI-installed CLI that runs offline and deterministically. It has no LLM, no cloud calls, and no network dependency. Every finding maps to a source-linked rule in an auditable catalog, and it goes deep on Python v2 runtime, hosting-plan, and deploy compatibility. It is safe to run unattended in CI and on pull-request workspaces.
• The official azure-functions-skills doctor — a multi-language coding-agent skill whose headline value is LLM semantic analysis. Its --deep mode runs an AI agent (with elevated file-write / shell-exec permissions) to catch semantic issues that deterministic rules cannot.

These are complementary. You can run this package as a fast, deterministic pre-deploy gate and use the official skill's AI analysis for semantic review inside your coding agent.

Comparison

Dimension	azure-functions-doctor (this package)	azure-functions-skills doctor (official)
Primary form	PyPI Python package + official GitHub Action	Coding-agent plugin (skills + Azure MCP), distributed via npm @azure/functions-skills and plugin marketplaces
Core engine	Deterministic rule catalog — no LLM required	LLM semantic analysis is the headline value (--deep); a --no-deep Tier-1 deterministic mode also exists
Offline / CI-safe	Yes — fully offline, no network or cloud, safe to run on PR workspaces	Deep mode needs an AI agent with elevated permissions and refuses to run on PR workspaces (prompt-injection risk); --no-deep is the CI-safe tier
Transparency / auditability	Every finding cites a rule in a source-linked catalog; output is reproducible	Deep findings come from LLM reasoning; --no-deep is deterministic
Language focus	Python v2 decorator model — deep runtime/hosting-plan/deploy checks (Python lifecycle, hosting-plan Python caps, Flex config, binding connection resolution)	Multi-language, not Python-specific
Runtime required	Python 3.10+	Node.js 20+ (Node 24+ for Copilot CLI)
Determinism	Deterministic — same inputs produce the same findings	Deep mode is non-deterministic; --no-deep is deterministic

Naming relationship

This project is an independent community package published on PyPI as azure-functions-doctor. It is not affiliated with, endorsed by, or maintained by Microsoft. The official skill lives at templates/skills/azure-functions-doctor inside Azure/azure-functions-skills and happens to share the doctor name because both diagnose Azure Functions projects.

To disambiguate:

• This package is what you pip install azure-functions-doctor and invoke as azure-functions-doctor doctor.
• The official skill is what you add to a coding agent via the azure-functions-skills plugin.

When to use which

• Reach for this package when you want a fast, offline, deterministic gate that runs the same way locally, in CI, and on pull requests — with every finding traceable to a documented rule, focused on Python v2 deploy compatibility.
• Reach for the official skill when you want AI-assisted semantic review (missing error handling, blocking I/O, hardcoded secrets, durable-orchestrator non-determinism) from inside your coding agent, and you can grant it a trusted workspace.

They are not mutually exclusive — running both gives you deterministic pre-deploy coverage plus AI-assisted semantic analysis.