macOS
Native releases for Apple Silicon and Intel, with the one-line installer or Homebrew.
aarch64 + x86_64 support
What Squeezy supports.
The platforms it runs on, the languages it understands, the models you can bring, and the fastest way to send useful, redacted context when something goes wrong.
platforms 3 macOS, Linux, Windows
languages 22 with local code understanding
providers 27 plus any OpenAI-compatible endpoint
platforms
Runs natively on macOS, Linux, and Windows.
On Windows, the default sandbox tier enforces filesystem-write isolation without admin; full read and network isolation is an opt-in tier you set up once. The permissions docs have the detail.
Linux
Static x86_64 and ARM64 binaries, plus installation from crates.io when the Rust toolchain is already present.
x86_64 + aarch64Windows
An x86_64 native release through Winget or a downloadable archive. The default sandbox enforces filesystem writes; full read and network isolation is opt-in.
x86_64 languages
22 languages get local code understanding.
For these languages, Squeezy navigates declarations, references, and structure on your machine. Files outside the list still work through bounded search, targeted reads, and verifier commands.
Rust
Cargo workspaces, crates, traits, impls, modules, and tests.
Python
Packages, imports, classes, functions, decorators, and inheritance.
Java
Packages, Maven/Gradle projects, classes, members, and inheritance.
Kotlin
Packages, Gradle projects, classes, objects, companions, and extensions.
Scala
Packages, traits, objects, case classes, enums, and extension methods.
Markdown
Headings, sections, links, symbol references, and documentation structure.
AsciiDoc
Section titles, anchors, inline references, and documentation structure.
reStructuredText
Underline sections, Sphinx cross-reference roles, and inline literals.
Plain Text
Indexed text file roots for navigation and search.
C#
Solutions, namespaces, usings, partial types, attributes, and members.
Go
Modules, packages, structs, interfaces, receivers, imports, and tests.
C
Headers, includes, structs, functions, typedefs, macros, and references.
C++
Headers, namespaces, classes, templates, methods, and overload-heavy code.
CSS
Selectors, classes, IDs, keyframes, and stylesheet structure.
HTML
Elements, IDs, classes, tags, and document structure.
JavaScript
ES modules, CommonJS, functions, classes, exports, and JSX.
TypeScript
Types, interfaces, imports, generics, classes, and TSX.
PHP
Namespaces, Composer-style code, traits, enums, attributes, and methods.
Ruby
Classes, modules, mixins, singleton methods, accessors, and require paths.
Shell
Scripts, functions, commands, and executable shell entry points.
Swift
SwiftPM modules with enforced visibility, protocols, actors, structs, extensions, and property wrappers.
Dart
Pubspec package identity, libraries, parts, classes, mixins, extensions, and named arguments in Flutter-style projects.
Full coverage and the per-language detail live in the language docs.
providers
Bring your own model.
Twenty-seven providers across aggregators, first-party APIs, subscription sign-ins, cloud platforms, open-weight hosts, and local runtimes, plus any OpenAI-compatible endpoint. Keys and credentials stay on your machine.
First-party & subscription
OpenAI, Anthropic, and Google Gemini by API key, plus supported subscription OAuth routes.
API key or OAuthAggregators & gateways
One key, many models: OpenRouter, Vercel AI Gateway, PortKey, and Cloudflare AI Gateway.
one key, many modelsCloud platforms & open-weight hosts
Cloud IAM via AWS Bedrock, Azure OpenAI, and Google Vertex; open-weight hosts Groq, Cerebras, DeepSeek, Together, Fireworks, DeepInfra, Baseten, Mistral, xAI, and Cloudflare Workers AI.
IAM or API keyLocal & custom
Run locally with Ollama, LM Studio, vLLM, or llama.cpp, or point the OpenAI-compatible preset at any other endpoint.
local or customExact environment variables and auth methods are in the provider docs.
diagnostics
Use the in-product reporting paths first.
These keep support tied to the session instead of a vague bug report. You can preview any report bundle before sending it.
/feedback
Best for short product feedback, wording issues, confusing UI, or a feature request that does not need a full session bundle.
/report
Best when the current run failed, produced a confusing answer, hit a provider issue, or needs a redacted diagnostic archive.
squeezy sessions report
Best when the issue happened earlier. Use --preview first, then send only if the redacted bundle looks acceptable.
/feedback
/report
squeezy sessions report <session_id> --preview
squeezy sessions report <session_id> --sendKeep the report id with the symptom.
If support asks for more detail, share the sanitized symptom, platform, provider route, the command that failed, and the feedback or report id. Do not paste secrets, private code, or report-archive contents into a message.