@su-record/vibe 3.2.42 → 3.2.43
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": "vibe",
|
|
3
3
|
"displayName": "Vibe",
|
|
4
|
-
"version": "3.2.
|
|
4
|
+
"version": "3.2.43",
|
|
5
5
|
"description": "Verification harness for AI coding agents — \"done\" is decided by deterministic gates (test exit codes, run-ledger, regression memory), not the model self-report.",
|
|
6
6
|
"author": {
|
|
7
7
|
"name": "su-record",
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "vibe",
|
|
3
|
-
"version": "3.2.
|
|
3
|
+
"version": "3.2.43",
|
|
4
4
|
"description": "Verification harness for AI coding agents — \"done\" is decided by deterministic gates (test exit codes, run-ledger, regression memory), not the model self-report.",
|
|
5
5
|
"author": {
|
|
6
6
|
"name": "su-record",
|
package/agents/e2e-tester.md
CHANGED
|
@@ -26,34 +26,18 @@ failed scenario. The loop, not a human, closes the gap.
|
|
|
26
26
|
|
|
27
27
|
Cheap verification is what makes the closed loop viable. Priority order:
|
|
28
28
|
|
|
29
|
-
1. **
|
|
30
|
-
|
|
31
|
-
|
|
32
|
-
```bash
|
|
33
|
-
npm install -g agent-browser && agent-browser install # native CLI (no Node/Playwright at runtime)
|
|
34
|
-
agent-browser mcp # or run it as an MCP server
|
|
35
|
-
```
|
|
36
|
-
|
|
37
|
-
CLI shape — snapshot once, then act on refs:
|
|
38
|
-
|
|
39
|
-
```bash
|
|
40
|
-
agent-browser open localhost:3000
|
|
41
|
-
agent-browser snapshot # accessibility tree with refs
|
|
42
|
-
agent-browser click @e2
|
|
43
|
-
agent-browser fill @e3 "user@example.com"
|
|
44
|
-
agent-browser close
|
|
45
|
-
```
|
|
46
|
-
|
|
47
|
-
2. **Playwright test runner** — write a spec, run
|
|
48
|
-
`npx playwright test --reporter=line`, consume pass/fail only.
|
|
29
|
+
1. **Playwright test runner** — write a spec, run
|
|
30
|
+
`npx playwright test --reporter=line`, consume **pass/fail only**. The spec
|
|
31
|
+
file holds the interaction detail so the transcript doesn't.
|
|
49
32
|
If missing: `npx playwright install chromium`.
|
|
50
|
-
|
|
51
|
-
context in 2–3
|
|
52
|
-
|
|
53
|
-
|
|
54
|
-
>
|
|
55
|
-
>
|
|
56
|
-
>
|
|
33
|
+
2. **Playwright MCP (DOM-based)** — interactive poking when a spec can't
|
|
34
|
+
express the check. Last resort: full DOM trees exhaust context in 2–3
|
|
35
|
+
interactions, so budget for it rather than defaulting to it.
|
|
36
|
+
|
|
37
|
+
> **비용이 루프 횟수를 정한다.** 검증 한 번이 비쌀수록 세션당 돌릴 수 있는 루프가
|
|
38
|
+
> 줄고, 루프가 줄면 닫힌 루프가 성립하지 않는다. 그래서 기본은 **결과만 소비하는**
|
|
39
|
+
> 1순위다 — 상호작용 내용은 spec 파일에 남기고 컨텍스트에는 exit code 만 들인다.
|
|
40
|
+
> DOM 을 컨텍스트로 끌어오는 방식은 그 예산을 한 번에 태운다.
|
|
57
41
|
|
|
58
42
|
## Verification Scope
|
|
59
43
|
|
package/package.json
CHANGED
|
@@ -10,13 +10,12 @@ Browser Tool Priority:
|
|
|
10
10
|
|
|
11
11
|
| Priority | Tool | 용도 | 없을 때 |
|
|
12
12
|
|----------|------|------|---------|
|
|
13
|
-
| 1st |
|
|
14
|
-
| 2nd | Playwright
|
|
15
|
-
| 3rd | Playwright MCP (DOM) | 최후 수단, 토큰 비효율 | — |
|
|
13
|
+
| 1st | Playwright Test Runner | 테스트 코드 실행, **pass/fail 만** 소비 | `npx playwright install chromium` |
|
|
14
|
+
| 2nd | Playwright MCP (DOM) | spec 으로 표현 못 하는 확인, 최후 수단 | — |
|
|
16
15
|
|
|
17
|
-
>
|
|
18
|
-
>
|
|
19
|
-
>
|
|
16
|
+
> 검증 비용이 루프 횟수를 정한다. 1순위는 상호작용 내용을 spec 파일에 남기고
|
|
17
|
+
> 컨텍스트에는 exit code 만 들인다. DOM 을 컨텍스트로 끌어오는 2순위는 2~3회면
|
|
18
|
+
> 예산을 태우므로 기본값이 아니라 예외로 쓴다.
|
|
20
19
|
|
|
21
20
|
**활성화 조건:** Feature 파일에 UI 관련 시나리오 존재 + `.vibe/e2e/config.json`의 `closedLoop.enabled: true` (기본값) + dev server가 실행 중
|
|
22
21
|
|