Install · QWED

Python

Requires Python 3.10 or newer. The base package installs the core engines: Math, Code and Facts.

shell

$ pip install qwed

For the SQL, Logic (Z3) and Symbolic (CrossHair) engines, install the full extra:

shell

$ pip install "qwed[full]"

TypeScript, Go and Rust

Language	Package	Install
TypeScript	`@qwed-ai/sdk`	`npm install @qwed-ai/sdk`
Go	`sdk-go`	`go get github.com/QWED-AI/qwed-verification/sdk-go`
Rust	`qwed`	`cargo add qwed`

Docker

The image is part of the Docker Sponsored Open Source program.

shell

$ docker pull qwedai/qwed-verification

From source

shell

$ git clone https://github.com/QWED-AI/qwed-verification.git$ cd qwed-verification$ pip install -e .

First run

qwed init starts the engines, runs a short verification suite, and then asks for your model provider and key.

shell

$ mkdir my-project && cd my-project$ qwed init[QWED] Initializing verification engines...  [ok] SymPy    math engine ready  [ok] Z3       logic engine ready  [ok] AST      code engine ready  [ok] SQLGlot  sql engine readyRunning verification suite...  [ok] 2+2=5                    -> BLOCKED  [ok] x>5 AND x<3              -> UNSAT  [ok] SELECT * WHERE 1=1       -> BLOCKED  [ok] eval(user_input)         -> BLOCKEDAll engines verified. QWED is operational.Step 1/3: Select your LLM provider (NVIDIA, OpenAI, Anthropic, Gemini, Custom)Step 2/3: Enter API key — tested with 5s timeout, stored securely (.env, 0600)Step 3/3: QWED API key generated — shown once, save itQWED is ready.

Then check the setup and run the built-in tests:

shell

$ qwed doctor$ qwed test

qwed test runs 12 deterministic tests. All of them should pass before you put QWED in front of production traffic.

Model providers

Pick a provider with --provider. Any OpenAI-compatible API works through custom.

shell

$ qwed init --provider nvidia$ qwed init --provider openai$ qwed init --provider anthropic$ qwed init --provider gemini$ qwed init --provider custom

In CI, skip the prompts:

shell

$ qwed init --non-interactive --provider nvidia# or with an environment variable$ NVIDIA_API_KEY=xxx qwed init --non-interactive

To run against a local model such as Ollama, see LLM configuration in the docs.

Domain packages

Each domain package adds guards for one field and installs on its own.

Package	Latest	Install
qwed-finance Banking, payments and financial calculations	3.0.1	`pip install qwed-finance`
qwed-legal Deadlines, amounts and legal claims	0.5.1	`pip install qwed-legal`
qwed-tax Tax compliance and withholding	0.2.1	`pip install qwed-tax`
qwed-infra Terraform, IAM and cloud cost	0.3.0	`pip install qwed-infra`
qwed-open-responses Agent tool calls, checked before they run	0.6.1	`pip install qwed-open-responses`
qwed-mcp MCP server for any MCP client	0.2.2	`pip install qwed-mcp`
qwed-a2a Agent-to-agent verification	0.3.0	`pip install qwed-a2a`
qwed-ucp Universal Commerce Protocol transactions	0.3.0	`pip install qwed-ucp`

Next: the quickstart in the docs Install problem? Open an issue