sniff-qa 0.9.1 → 0.9.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -1,7 +1,7 @@
1
1
  {
2
2
  "name": "sniff",
3
- "version": "0.9.1",
4
- "description": "Autonomous QA scanner via sniff-qa: walks your running app in a real browser and reports real bugs (broken pages/links, console/network errors, broken forms, state-loss, empty/placeholder data, bad loading/error states, responsive + a11y) with reproduction proof, severity, confidence, and a fix. No API key.",
3
+ "version": "0.9.2",
4
+ "description": "Walks your running app in a real browser and lists what is broken: dead pages, console errors, broken forms. Each page bug comes with repro steps and a fix.",
5
5
  "author": {
6
6
  "name": "Adam Boudjemaa",
7
7
  "url": "https://github.com/Aboudjem/sniff"
package/CHANGELOG.md CHANGED
@@ -7,6 +7,23 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
7
7
 
8
8
  ## [Unreleased]
9
9
 
10
+ ## [0.9.2] - 2026-11-17
11
+
12
+ A clearer first screen and truer wording. No behaviour change: the code is the same as 0.9.1.
13
+
14
+ ### Changed
15
+
16
+ - **README first screen.** It now opens with the outcome, the command and the real report. Under
17
+ the report, a short `Why use it` block gives three benefits, each with its proof. The limit sits
18
+ next to the promise: a walk submits real forms, so use a dev or staging app.
19
+ - **Listing descriptions.** The plugin, npm, Gemini and MCP registry descriptions are now one plain
20
+ sentence instead of a feature list.
21
+ - **Truer wording.** The README no longer says every finding has a screenshot. On the fixture, 23 of
22
+ 27 findings have one, and source-scan findings have no route. The first-run size is now measured:
23
+ on a Mac, `npx sniff-qa install` left 193 MB of engine and 528 MB of Chromium on disk.
24
+ - **Translations, hero banners, social preview, `llms.txt`, `AGENTS.md` and the Pages site** carry
25
+ the same one-liner.
26
+
10
27
  ## [0.9.1] - 2026-11-17
11
28
 
12
29
  The MCP server now connects on a new machine's first start.
package/README.md CHANGED
@@ -12,7 +12,7 @@
12
12
 
13
13
  <p align="center"><b>English</b> · <a href="READMEs/zh-CN.md">简体中文</a> · <a href="READMEs/ja.md">日本語</a> · <a href="READMEs/es.md">Español</a> · <a href="READMEs/fr.md">Français</a></p>
14
14
 
15
- <p align="center"><b>Point it at your running app. It clicks through your pages in a real browser and tells you what is broken, with the steps to reproduce it.</b></p>
15
+ <p align="center"><b>Catch what is broken in your web app before your users do.</b></p>
16
16
 
17
17
  <p align="center"><a href="#what-you-get">What you get</a> · <a href="#install">Install</a> · <a href="#how-it-works">How it works</a> · <a href="#works-with">Works with</a> · <a href="#proof">Proof</a> · <a href="#faq">FAQ</a> · <a href="#limits">Limits</a></p>
18
18
 
@@ -24,7 +24,6 @@ npx sniff-qa --url http://localhost:3000
24
24
 
25
25
  A report that opens with the verdict, then the three things to fix first, then every finding.
26
26
  This is the report from a real run on this repo's planted-bug fixture (2026-11-17, sniff 0.9.0).
27
- The progress lines that print before it are left out.
28
27
 
29
28
  ```text
30
29
  VERDICT FAIL: 26 findings (1 critical, 13 high, 11 medium, 1 low)
@@ -46,27 +45,26 @@ The progress lines that print before it are left out.
46
45
  shot: sniff-reports/crawl/_crash-desktop.png
47
46
  ```
48
47
 
49
- Each finding carries a severity, a confidence label, the route, the steps, a fix and a screenshot.
48
+ You changed one form, and now a page you never opened throws a 500. sniff opens your running app
49
+ in a real browser and clicks through it like a user would. Each page bug it reports comes with
50
+ the steps that reproduce it.
50
51
 
51
- To reproduce it from a clone, start the fixture, then walk it from a second terminal:
52
+ ### Why use it
53
+
54
+ - **Fix it, do not hunt for it.** Each page finding names the route, the steps and a fix, most with a screenshot.
55
+ - **Proof it catches real bugs.** Tested on bugs planted on purpose: 21 of 21 found, 0 false findings on the clean page (`npm run build && npm run proof`).
56
+ - **Block bad merges.** `--fail-on` exits 1 on critical or high findings, and `npx sniff-qa ci` writes the GitHub Actions file. No API key.
57
+
58
+ It needs your app running, and it submits real forms, so use a dev or staging app.
59
+ The first walk downloads a browser once, about 720 MB on disk on a Mac.
60
+
61
+ To reproduce the sample from a clone, start the fixture, then walk it from a second terminal:
52
62
 
53
63
  ```bash
54
64
  node sniff-tests/planted-bugs/server.mjs 4517
55
65
  npx sniff-qa --url http://localhost:4517
56
66
  ```
57
67
 
58
- | Output | Flag | Use it for |
59
- | :-------------- | :---------- | :------------------------------------------ |
60
- | Terminal report | (default) | Reading the result now |
61
- | HTML page | `--report` | Sending to someone, one self-contained file |
62
- | JSON | `--json` | Scripts and other tools |
63
- | Exit code | `--fail-on` | Failing CI on real bugs |
64
-
65
- | Exit code | Meaning |
66
- | :-------- | :------------------------------------------------------------------------------ |
67
- | `0` | No finding reached the `--fail-on` severities (default: critical and high) |
68
- | `1` | A finding reached them, an `assert` budget was breached, or sniff could not run |
69
-
70
68
  <img alt="sniff walking a buggy app and streaming findings with severity, confidence, steps to reproduce, and a fix" src="https://raw.githubusercontent.com/Aboudjem/sniff/main/.github/assets/demo.gif" width="100%">
71
69
 
72
70
  ## Install
@@ -101,18 +99,42 @@ claude mcp add sniff-qa -- npx -y sniff-qa --mcp
101
99
  ```
102
100
 
103
101
  The first walk downloads two things, once: the browser engine (Playwright, axe-core and
104
- Lighthouse, about 200 MB) and Chromium (about 165 MB). Both took 159 s here on 2026-11-17.
105
- To do it ahead of time, run:
102
+ Lighthouse) and Chromium. To do it ahead of time, run:
106
103
 
107
104
  ```bash
108
105
  npx sniff-qa install
109
106
  ```
110
107
 
111
- It ends with the time it took. `npx sniff-qa doctor` checks Node, the browser and your dev server.
108
+ It ends with the time it took. On a Mac with an empty npm cache it printed 34 s (2026-11-17, sniff 0.9.1).
109
+ It left 193 MB of engine and 528 MB of Chromium on disk. To re-measure on macOS:
110
+
111
+ ```bash
112
+ du -sm ~/.cache/sniff-qa/engine/* ~/Library/Caches/ms-playwright
113
+ ```
112
114
 
113
- The MCP server starts before those downloads. On a cold machine (empty npm cache),
114
- `npx -y sniff-qa@0.9.1 --mcp` answered its first handshake in 1.4 to 1.9 s (5 runs of the release build). Version 0.9.0 took 23 to 92 s,
115
- past Claude Code's 30 s connect limit. `npm run cold-start` re-measures it in a clone of this repo.
115
+ `npx sniff-qa doctor` checks Node, the browser and your dev server.
116
+
117
+ The MCP server starts before those downloads. With an empty npm cache, `npx -y sniff-qa@0.9.1 --mcp`
118
+ answered its first handshake in 3.1 to 4.9 s (5 runs, 2026-11-17). Version 0.9.0 took 37 to 93 s,
119
+ past Claude Code's 30 s connect limit. To re-measure, in a clone of this repo:
120
+
121
+ ```bash
122
+ npm run cold-start -- 0.9.1 5
123
+ ```
124
+
125
+ A walk prints progress lines, then the report. You can take the result four ways:
126
+
127
+ | Output | Flag | Use it for |
128
+ | :-------------- | :---------- | :------------------------------------------ |
129
+ | Terminal report | (default) | Reading the result now |
130
+ | HTML page | `--report` | Sending to someone, one self-contained file |
131
+ | JSON | `--json` | Scripts and other tools |
132
+ | Exit code | `--fail-on` | Failing CI on real bugs |
133
+
134
+ | Exit code | Meaning |
135
+ | :-------- | :------------------------------------------------------------------------------ |
136
+ | `0` | No finding reached the `--fail-on` severities (default: critical and high) |
137
+ | `1` | A finding reached them, an `assert` budget was breached, or sniff could not run |
116
138
 
117
139
  <details>
118
140
  <summary>Project install and CI</summary>
@@ -128,7 +150,7 @@ with browser caching and report artifacts.
128
150
  1. **Find the app.** It uses `--url`, or a dev server on a common port.
129
151
  2. **Walk it.** A headless Chromium follows your links, clicks buttons and fills forms, like a user.
130
152
  3. **Judge what happened.** Each page is checked for 12 classes of bugs.
131
- 4. **Prove it.** A finding needs a route, the steps, a screenshot and the console or network excerpt.
153
+ 4. **Prove it.** A finding needs a route and the steps. It adds a screenshot and the console or network lines when it has them.
132
154
  5. **Report.** Verdict first, then the fix-first list, then the detail.
133
155
 
134
156
  The 12 classes:
@@ -229,7 +251,7 @@ More answers are in [docs/faq.md](docs/faq.md).
229
251
  ## Limits
230
252
 
231
253
  - It needs a running app for the walk. Without one you only get the source scan.
232
- - The first walk needs internet to download the browser engine and Chromium once, about 365 MB.
254
+ - The first walk needs internet to download the browser engine and Chromium once, 721 MB on disk on a Mac.
233
255
  - Until then, MCP browser tools answer `needsSetup` and the agent calls `sniff_install` first.
234
256
  - Dead-link checking follows external links, so a walk requests the third-party URLs your pages link to.
235
257
  - Redaction covers report text, not screenshot pixels. A screenshot of a logged-in page can show private data.
package/package.json CHANGED
@@ -1,7 +1,7 @@
1
1
  {
2
2
  "name": "sniff-qa",
3
- "version": "0.9.1",
4
- "description": "Autonomous QA scanner: point it at your running app and it walks the real user flows in a headless browser and reports what's actually broken: broken pages/links, console and network errors, empty/placeholder data, broken forms, state-loss, bad loading/error states, responsive and accessibility issues, each with reproduction proof, severity, confidence, and a fix. CLI + MCP server. No API key. Uses Playwright and axe-core.",
3
+ "version": "0.9.2",
4
+ "description": "Walks your running app in a real browser and lists what is broken: dead pages, console errors, broken forms. Each page bug comes with repro steps and a fix.",
5
5
  "type": "module",
6
6
  "main": "./dist/index.js",
7
7
  "types": "./dist/index.d.ts",