npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

@qa-ai-stlc/mcp-server

v1.16.0

Published

Local stdio MCP server over the core engine API: a tool registry with Zod input/output schemas and coded errors. Calls no model itself.

Readme

@qa-ai-stlc/mcp-server

A local stdio MCP server over @qa-ai-stlc/core: a tool registry with Zod input/output schemas (AGENTS.md 12.3). A ToolDefinition's handler returns a value matching its outputSchema or throws — a QaError for an expected failure, anything else for a bug — and registerTool turns that into a structured CallToolResult carrying a stable code and, when there is one, a remediation, never a raw stack trace. The server calls no model itself; it is driven by the agent host's own model over the MCP protocol.

qa-mcp-server starts the stdio server with the built-in tools: qa.ping (a health check that does not touch .qa/) and one tool per engine operation — qa.doctor, qa.init, qa.config_show, qa.config_set, qa.config_add, qa.explore, qa.api_diff, qa.scope, qa.scope_set, qa.cases_add, qa.cases_render, qa.approve, qa.defect_add, qa.defect_accept, qa.rca_input, qa.rca_add, qa.rca_approve, qa.validate, qa.run, qa.report — each calling the exact same packages/core or packages/explorer function its CLI counterpart calls (P2-05). qa.explore does not support manual pick-mode capture: opening a headed browser for a human to click through is a CLI-only feature, not something an agent can drive over stdio.

Interactive case execution (P3-14/P3-15) adds qa.http_execute (the api test type, no browser), qa.registry_execute_register and qa.case_result_register. qa.generation_proven_session (P3-07) reads back a case's most recently proven qa-execute session — its evidence grouped by the stepId each browser.*/qa.http_execute call tagged it with — for qa-generate-tests to codify into a spec.

It also starts the six browser.* tools — qa.browser_open, qa.browser_navigate, qa.browser_click, qa.browser_fill, qa.browser_snapshot, qa.browser_close — the only way an agent may touch the application under test (ADR-005). Every one of them registers what it did as hashed evidence under the session's run before it returns, so an exploratory session leaves a complete trail and an unregistered screenshot is not a representable outcome. A session runs in safe mode with no opt-out: requests that are not safe reads (GET, HEAD, OPTIONS, QUERY) are aborted, and navigation is limited to the environment's domain allowlist. For component library widgets (ui.componentLibrary) four more tools act at the widget level instead of replaying clicks: qa.browser_select_option (an option by its visible text), qa.browser_set_date, qa.browser_open_popup and qa.browser_close_popup. Each takes the selector the registry holds for the widget, finds the widget wrapper from it, and verifies the widget's resulting state before it registers the evidence: a widget that does not show the choice fails with BROWSER_WIDGET_VALUE_MISMATCH, which can be a defect worth reporting. For data grids, qa.browser_grid_find_row finds a row by a column header and a cell value, paging or scrolling through a grid that does not render every row, and qa.browser_grid_read_cell reads one cell of that row; a row the grid never shows fails with BROWSER_GRID_ROW_NOT_FOUND. qa.browser_attach is the other way to get a session: it connects to a Chrome the operator started with --remote-debugging-port and already signed into (SSO, MFA), for an application the engine cannot sign into itself. The endpoint must be on this machine; the session drives the first open page that is on the environment allowlist, under safe mode, and closing it disconnects without closing the operator's browser. Unlike every other tool, these share one session store owned by the server process, because a session spans several tool calls; it is closed on SIGINT/SIGTERM, and an idle session is closed at the next call that touches it.

The project the tools act on is the nearest directory from the server's working directory upward that holds .qa/, or the working directory itself. QA_PROJECT_ROOT (an absolute path to an existing directory) overrides that. A server started inside its plugin's install directory (PLUGIN_ROOT) fails with PROJECT_ROOT_UNRESOLVED instead of acting on it, because Codex starts a plugin-declared server there; register the server outside the plugin so it starts in the project.

Part of QA-AI-STLC. See the repository root for license, contributing and security information.