AI-powered project scanner with TUI output and optional automated fixes. Scanner uses a set of tools defined by the user to run various automated tests. After running these tests, scanner launches an AI solver per failed check to diagnose and fix problems.

Scanner requires an AI coding agent to analyze and fix errors. You must have one of the following installed and configured:

# Install Codex CLI
npm install -g @openai/codex

# Login and configure
codex auth

# Install Claude Code
npm install -g @anthropic-ai/claude-code

# Login and configure
claude login

Once configured, specify which agent to use:

• Via CLI: scanner --agent codex or scanner --agent claude
• Via config: Set [agent] in scanner.toml

All check tools are expected to return results in GitHub Actions annotation format (::error file=X,line=Y::message). If your tool outputs a different format (e.g., JSON), specify a formatter command that converts the output to GHA format.

curl -sSL https://raw.githubusercontent.com/getresq/scanner/master/scripts/install.sh | bash

Install Rust toolchain (1.79+), cargo.

cargo build --release
# binary: target/release/scanner

To install into your toolchain’s bin dir:

cargo install --path .

• Run scanner: scanner
• TUI keys: ↑/↓ move, y copy details, q/esc exit (double-press while checks run).

See scanner.toml for checks and agent settings. Each project can keep its own config alongside the codebase.

Tip: if multiple checks contend for a shared resource (package manager cache, codegen outputs, etc.), set a shared lock = "name" on those checks to force them to run one-at-a-time while keeping other checks concurrent.