@vostride/agent-qa-model 0.2.2 → 0.2.4

package/README.md

@@ -0,0 +1,78 @@
+<picture>
+  <source media="(prefers-color-scheme: dark)" srcset="https://raw.githubusercontent.com/vostride/agent-qa/main/docs/assets/readme/banner-dark.jpg">
+  <source media="(prefers-color-scheme: light)" srcset="https://raw.githubusercontent.com/vostride/agent-qa/main/docs/assets/readme/banner-light.png">
+  <img src="https://raw.githubusercontent.com/vostride/agent-qa/main/docs/assets/readme/banner-light.png" alt="agent-qa by vostride — a self-improving QA agent for your software factory" width="100%">
+</picture>
+
+<p align="center">
+  <strong>English</strong> · <a href="https://github.com/vostride/agent-qa/blob/main/README.zh-CN.md">中文</a>
+</p>
+
+<p align="center">
+  <a href="https://www.npmjs.com/package/agent-qa"><img src="https://img.shields.io/npm/v/agent-qa?color=196872" alt="npm version"></a>
+  <a href="https://www.npmjs.com/package/agent-qa"><img src="https://img.shields.io/npm/dm/agent-qa?color=196872" alt="npm downloads per month"></a>
+  <a href="https://github.com/vostride/agent-qa/stargazers"><img src="https://img.shields.io/github/stars/vostride/agent-qa?color=196872" alt="GitHub stars"></a>
+  <a href="https://github.com/vostride/agent-qa/actions/workflows/ci.yml"><img src="https://github.com/vostride/agent-qa/actions/workflows/ci.yml/badge.svg?branch=main" alt="CI"></a>
+  <a href="https://vostride.com/docs/agent-qa"><img src="https://img.shields.io/badge/docs-agent--qa-196872" alt="Documentation"></a>
+  <a href="https://github.com/vostride/agent-qa/blob/main/LICENSE.md"><img src="https://img.shields.io/badge/license-FSL--1.1--ALv2-196872" alt="License: FSL-1.1-ALv2"></a>
+</p>
+
+<p align="center">
+  <a href="https://vostride.com/agent-qa">Website</a> · <a href="https://vostride.com/docs/agent-qa">Docs</a> · <a href="https://vostride.com/demo/agent-qa">Demo</a> · <a href="https://vostride.com/blog">Blog</a> · <a href="https://vostride.com/changelog">Changelog</a>
+</p>
+
+**Your AI QA engineer, with memory.** Describe a user journey in plain language. agent-qa tests it on web, Android or iOS and keeps the evidence. Use the CLI, the dashboard, or your coding agent through MCP and skills.
+
+- **memidex:** remembers facts and working procedures as reviewable files in your repository.
+- **Self-improvement:** validates lessons from runs, retires stale knowledge, and uses your feedback to correct mistakes.
+
+| Benchmark | Rank | Published result |
+| --- | --- | --- |
+| [AndroidWorld](https://vostride.com/benchmark/android-world) | **#1** | **100% pass@1** (116/116 tasks) |
+| [AppControlBench](https://vostride.com/benchmark/app-control-bench) | **#1** | **99.6% completion** (114.5/115 points) |
+
+Read the full benchmark reports, with run evidence and methodology, on [our blog](https://vostride.com/blog). Both used gpt-6-sol with memory and self-improvement off; AppControlBench gives half credit for partial completion.
+
+## See it in action
+
+**Inspect a failure.** The demo catches a cart showing $29 instead of $53, with the failed step and its screenshot side by side.
+
+<picture>
+  <source media="(prefers-color-scheme: dark)" srcset="https://raw.githubusercontent.com/vostride/agent-qa/main/docs/assets/readme/evidence-dark.png">
+  <source media="(prefers-color-scheme: light)" srcset="https://raw.githubusercontent.com/vostride/agent-qa/main/docs/assets/readme/evidence-light.png">
+  <img src="https://raw.githubusercontent.com/vostride/agent-qa/main/docs/assets/readme/evidence-light.png" alt="The dashboard shows the deliberate cart-total bug, its failed assertion, and recorded browser evidence" width="100%">
+</picture>
+
+**See what memory contributed.** Inspect the knowledge supplied to a run and trace it back to the steps that used it.
+
+<picture>
+  <source media="(prefers-color-scheme: dark)" srcset="https://raw.githubusercontent.com/vostride/agent-qa/main/docs/assets/readme/knowledge-dark.png">
+  <source media="(prefers-color-scheme: light)" srcset="https://raw.githubusercontent.com/vostride/agent-qa/main/docs/assets/readme/knowledge-light.png">
+  <img src="https://raw.githubusercontent.com/vostride/agent-qa/main/docs/assets/readme/knowledge-light.png" alt="The Knowledge tab shows stored facts retrieved for the demo run" width="100%">
+</picture>
+
+**Run from your terminal.** Real output from the demo suite: checkout passes; the deliberate cart-total bug fails.
+
+<picture>
+  <source media="(prefers-color-scheme: dark)" srcset="https://raw.githubusercontent.com/vostride/agent-qa/main/docs/assets/readme/cli-dark.png">
+  <source media="(prefers-color-scheme: light)" srcset="https://raw.githubusercontent.com/vostride/agent-qa/main/docs/assets/readme/cli-light.png">
+  <img src="https://raw.githubusercontent.com/vostride/agent-qa/main/docs/assets/readme/cli-light.png" alt="CLI output excerpt showing one passed test, one failed test, and eight of nine steps passing" width="100%">
+</picture>
+
+## Get started
+
+Requires Node.js 24+.
+
+```sh
+npm install -D agent-qa
+npx agent-qa init
+```
+
+Follow the [quickstart](https://vostride.com/docs/agent-qa/quickstart) to connect a model and run the demo, or [set up with your coding agent](https://vostride.com/docs/agent-qa/agent-quickstart).
+
+> [!NOTE]
+> Early stage. Expect bugs. [Report an issue](https://github.com/vostride/agent-qa/issues) when something breaks.
+
+## License
+
+[Functional Source License (FSL-1.1-ALv2)](https://github.com/vostride/agent-qa/blob/main/LICENSE.md).

package/dist/index.cjs

@@ -790,7 +790,7 @@ var LOCAL_OPENAI_WIRE = [
   {
     id: "deepseek-chat",
     label: "DeepSeek Chat",
-    notes: "text-only model; use accessibility tree and DOM context."
+    notes: "Text-only model; use the accessibility tree."
   },
   { id: "deepseek-r1", label: "DeepSeek R1", notes: "Text-only reasoning model." },
   { id: "qwen2.5-vl", label: "Qwen2.5 VL" },
@@ -1326,10 +1326,7 @@ function seedBuiltins() {
     subscription: true,
     modelAdapter: "openai-responses",
     defaultModel: "gpt-6.1-sol",
-    // RUNX-01: the codex/ChatGPT-subscription Responses backend rejects the output cap
-    // (`max_output_tokens`) with HTTP 400. Strip it at egress for this provider ONLY —
-    // standard OpenAI-compatible `.chat()` endpoints and Anthropic support the cap.
-    // 2026-10-20 (446-11, D-18): it also answers 400 "Unsupported parameter: temperature".
+    // The subscription backend rejects `max_output_tokens` and `temperature` with HTTP 400.
     unsupportedParams: ["maxOutputTokens", "temperature"]
   });
   registerProvider({
@@ -1479,9 +1476,9 @@ var MIDDLEWARE = [
   toolSchemaMiddleware,
   deriveExamplesMiddleware(),
   (0, import_ai4.addToolInputExamplesMiddleware)(),
-  // RUNX-04: flag-gated debug logging (AGENT_QA_DEBUG_LLM); catches the raw APICallError first.
+  // Flag-gated by `AGENT_QA_DEBUG_LLM`; first, so it catches the raw `APICallError`.
   debugLoggingMiddleware(),
-  // RUNX-01: strip empty-content messages just before the wire.
+  // Strips empty-content messages just before the wire.
   dropEmptyMessagesMiddleware()
 ];
 function createModel(profile, auth, mask, overrides = {}, observation = {}) {
@@ -1630,7 +1627,7 @@ function resolveVisionConfig(servicesVision) {
 // src/vision/resolve-gate.ts
 var TEXT_ONLY_PROMPT_NOTE = [
   "Visual screenshot is unavailable for this step.",
-  "Rely on the accessibility tree and DOM context for element identification, including roles, names, values, refs, and URLs.",
+  "Rely on the accessibility tree for element identification, including roles, names, values, refs, and URLs.",
   "Verify functional state from the available structured context rather than purely visual appearance."
 ].join(" ");
 function visionFromConfig(configured, capabilities) {