@syedmuhammadali-dev/ai-qa-agent

Local agent CLI for the AI QA Agent platform. It runs on your machine, pairs with a project on the dashboard, and performs real analysis, testing, browser, accessibility, performance, API, architecture, and security checks against your codebase — reporting results back to the dashboard.

This is a standalone CLI — install it globally (or via npx) on any machine, in any project. It does not need to live inside, or be added as a dependency of, the ai-qa-agent monorepo itself; it talks to your dashboard over HTTP using the --url flag (or the URL you pass to connect).

Install

npm install -g @syedmuhammadali-dev/ai-qa-agent

Or run it directly without installing:

npx @syedmuhammadali-dev/ai-qa-agent status

Playwright is a runtime dependency and downloads its own Chromium binary during install (used by browser-check, a11y-check, and perf-check). If that postinstall step gets skipped for any reason, run it manually:

npx playwright install chromium

Quick start

# 1. Get a pairing code from your project's Settings page on the dashboard,
#    then pair this machine with that project:
ai-qa-agent connect <code> --url https://your-dashboard.example.com

# 2. Confirm it's connected:
ai-qa-agent status

# 3. Point it at a project on disk and run a check, e.g.:
ai-qa-agent analyze --cwd ./my-project
ai-qa-agent security-scan --cwd ./my-project

The pairing session is saved locally (~/.ai-qa-agent/session.json), so you only need to run connect once per machine per project — every other command reuses that session automatically.

Commands

ai-qa-agent <command> [options]

Connection

| Command | What it does | | --- | --- | | connect <code> [--url <apiUrl>] | Pair this machine with a project using a code generated in the dashboard's Settings page | | status | Show the current connection | | disconnect | Forget the saved connection on this machine |

ai-qa-agent connect <code> --url https://your-dashboard.example.com
ai-qa-agent status
ai-qa-agent disconnect

Running commands

| Command | What it does | | --- | --- | | run <command...> [--reason <text>] [--cwd <dir>] | Run an arbitrary command through the command policy engine, with audit logging | | analyze [--cwd <dir>] | Detect framework, package manager, test framework, DB/ORM, and auth from real files | | test [--cwd <dir>] | Detect the project's test setup and run it through the policy engine |

ai-qa-agent run npm run lint -- --reason "Check lint before merge"
ai-qa-agent analyze --cwd ./my-project
ai-qa-agent test --cwd ./my-project

Browser & web checks

| Command | What it does | | --- | --- | | browser-check <url> | Navigate to a URL with a real browser; reports console errors, failed requests, and a screenshot | | a11y-check <url> | Run the real axe-core accessibility engine against a URL | | perf-check <url> | Real browser Navigation/Resource Timing metrics for a URL |

ai-qa-agent browser-check https://example.com
ai-qa-agent a11y-check https://example.com
ai-qa-agent perf-check https://example.com

API & code checks

| Command | What it does | | --- | --- | | api-check <baseUrl> [--max-endpoints <n>] | Discover an OpenAPI spec (if any) and probe endpoints with real HTTP requests | | architecture [--cwd <dir>] | Analyze real import structure: circular deps, oversized files, coupling | | security-scan [--cwd <dir>] [--skip-audit] | Static secret/injection/XSS/CORS scan plus a real dependency audit |

ai-qa-agent api-check https://api.example.com --max-endpoints 20
ai-qa-agent architecture --cwd ./my-project
ai-qa-agent security-scan --cwd ./my-project

Fixes & releases

| Command | What it does | | --- | --- | | apply-fixes [--cwd <dir>] | Apply AI-proposed fixes approved on the dashboard, then run regression tests | | push-release [--cwd <dir>] | Push a release branch confirmed on the dashboard's pre-push screen (real branch, commit, push, PR — never main/master) | | export-zip [--cwd <dir>] [--out <dir>] | Export a clean, verified zip of the project: secret scan → cleanup → validate → README → integrity checksum |

ai-qa-agent apply-fixes --cwd ./my-project
ai-qa-agent push-release --cwd ./my-project
ai-qa-agent export-zip --cwd ./my-project --out .ai-qa/exports

Run ai-qa-agent --help or ai-qa-agent <command> --help for the full list of options for any command.

Requirements

• Node.js >= 20

License

MIT