@su-record/vibe 3.2.39 → 3.2.40

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": "3.2.39",
4
+ "version": "3.2.40",
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": "3.2.39",
3
+ "version": "3.2.40",
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",
@@ -26,15 +26,34 @@ 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. **Agent Browser MCP** — drives the browser via the accessibility tree
30
- (`button "Sign In"` 15 chars vs 200+ for the equivalent DOM). Dozens of
31
- loop iterations per session.
29
+ 1. **Agent Browser** — drives the browser via the accessibility tree, so a
30
+ control costs a ref (`@e2`) instead of a DOM subtree. Two ways in:
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
+
32
47
  2. **Playwright test runner** — write a spec, run
33
48
  `npx playwright test --reporter=line`, consume pass/fail only.
49
+ If missing: `npx playwright install chromium`.
34
50
  3. **Playwright MCP (DOM-based)** — last resort; full DOM trees exhaust
35
51
  context in 2–3 interactions.
36
52
 
37
- If Playwright is needed and missing: `npx playwright install chromium`.
53
+ > **1순위를 쓰려면 설치돼 있어야 한다.** 이전 판은 Agent Browser 를 1순위로
54
+ > 지정해 놓고 **얻는 방법을 적지 않았다** — 결과적으로 전원이 2순위로 떨어졌다.
55
+ > 설치 여부를 먼저 확인하고(`agent-browser --version`), 없으면 위 명령을 제안한
56
+ > 뒤 사다리를 내려간다. 사다리는 선언이 아니라 실행 가능해야 의미가 있다.
38
57
 
39
58
  ## Verification Scope
40
59
 
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@su-record/vibe",
3
- "version": "3.2.39",
3
+ "version": "3.2.40",
4
4
  "description": "AI Coding Framework for Claude Code — 7+ agents, 52 skills, multi-LLM orchestration",
5
5
  "type": "module",
6
6
  "main": "dist/cli/index.js",
@@ -8,11 +8,15 @@
8
8
 
9
9
  Browser Tool Priority:
10
10
 
11
- | Priority | Tool | 용도 |
12
- |----------|------|------|
13
- | 1st | Agent Browser (접근성 트리) | AI 직접 조작, 최소 토큰 |
14
- | 2nd | Playwright Test Runner | 테스트 코드 실행, pass/fail 반환 |
15
- | 3rd | Playwright MCP (DOM) | 최후 수단, 토큰 비효율 |
11
+ | Priority | Tool | 용도 | 없을 때 |
12
+ |----------|------|------|---------|
13
+ | 1st | Agent Browser (접근성 트리) | AI 직접 조작, 최소 토큰 | `npm i -g agent-browser && agent-browser install` |
14
+ | 2nd | Playwright Test Runner | 테스트 코드 실행, pass/fail 반환 | `npx playwright install chromium` |
15
+ | 3rd | Playwright MCP (DOM) | 최후 수단, 토큰 비효율 | — |
16
+
17
+ > 1순위는 **설치돼 있어야 1순위다.** 이전 판은 Agent Browser 를 1순위로 지정해
18
+ > 놓고 얻는 방법을 어디에도 적지 않아, 실제로는 전원이 2순위로 떨어졌다.
19
+ > `agent-browser --version` 으로 확인하고 없으면 위 명령을 제안한 뒤 내려간다.
16
20
 
17
21
  **활성화 조건:** Feature 파일에 UI 관련 시나리오 존재 + `.vibe/e2e/config.json`의 `closedLoop.enabled: true` (기본값) + dev server가 실행 중
18
22