pyre-check.org

What It Is

Pyre is a static type checker for Python, with a focus on gradual typing, incremental checking, and performance on large codebases. It can surface type errors in the terminal or editor, and follows typing standards such as PEP 484, 526, and 612. The site also notes that “Pyrefly” is the next generation of Pyre, but the crawled content is still mainly about Pyre and Pysa.

Core Capabilities

In terms of functionality, Pyre is well suited for gradually introducing type checking into existing Python projects: after installing pyre-check, run pyre init to generate a configuration, then execute pyre to detect issues such as mismatches with declared variable types. The first run starts a daemon that watches for file changes, and subsequent checks can be completed incrementally. Pysa is its security-focused static analysis component, which tracks data flow through models such as sources, sinks, rules, and sanitizers—for example, user-controlled input flowing into dangerous locations related to eval, SQL, or XSS.

Language, Integrations, and Documentation

The text clearly states support for Python, requires Python 3.9+, and depends on Watchman. Examples cover modeling approaches in ecosystems such as Django, sqlite3, and urllib. For integrations, the documentation mentions real-time feedback in the terminal and editors, and also provides Integrations and Editor Integration sections. The documentation quality is solid: beyond getting-started installation and configuration, it explains Pysa’s taint.config, .pysa models, rules, sanitizers, and common sources of false positives in detail.

Pricing and Deployment

The crawled text does not mention commercial pricing, a SaaS plan, or paid support. Usage is local: install via pip and run it inside the project, so it can be regarded as a locally self-hosted development tool. Whether it is open source is not explicitly stated in the text; the only confirmed point is that the site provides a GitHub link.

Pros, Cons, and Who It’s For

Its strengths are its focus on Python, support for large codebases and incremental checking, and the way it combines type checking with security taint analysis in a single toolchain. Downsides include the Watchman environment dependency and the relatively high maintenance cost of Pysa models; it also will not automatically see dependency source code that is not included in search_path, which may lead to false positives. It is suitable for Python platform teams, security engineering teams, and large projects that need long-term code quality governance. There is no information in the text about access from China, so it should be tested in practice. Alternatives to consider include mypy, Pyright, Semgrep, and Bandit.