v1.5.0 — AI-powered diagnosis

What's new

--ai flag — LLM-powered diagnosis for every failure and warning

Instead of just showing a static fix suggestion, --ai reads your actual
file contents and exact version numbers, sends them to an LLM, and streams
a specific, reasoned explanation back in real time.

ANTHROPIC_API_KEY=your-key npx predeploy-check --ai

Example — instead of a generic hint like "pin to Python 3.12", you get:

pydantic-core==2.20.0 predates Python 3.13's stable wheel infrastructure
— the cp313 wheel wasn't added until 2.23.0. Your fastest fix is bumping
to pydantic-core>=2.23.0. If you can't upgrade, pin runtime.txt to
python-3.12.7 instead.

Supports:

• Claude (claude-sonnet-4-6) via ANTHROPIC_API_KEY — preferred
• GPT-4o via OPENAI_API_KEY — fallback
• Requires Node.js 18+ (uses native fetch)
• Gracefully skipped if no API key is set, network is unavailable, or
  request times out — the tool always works without a key

Also in this release

Expanded npm keywords — bumped from 8 to 24 keywords covering
railway, nixpacks, nextjs, python, ci-cd, deployment-check,
static-analysis and more, for better discoverability in npm search.

Testing

Added 13 new tests for the AI module (113 total). AI tests use mocked
API calls so the full suite runs offline with no API key required.

Upgrade

npx predeploy-check@latest --help

The --ai flag shows its current status (which provider is active, or
how to enable it) directly in the help output.

Full Changelog: https://github.com/Alok-Fusion/predeploy_check/commits/v1.5.0