peaks-loop 4.0.40 → 4.0.41
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.
- package/CHANGELOG.md +14 -0
- package/README-en.md +1 -1
- package/README.md +1 -1
- package/package.json +5 -5
- package/skills/bee/peaks-perf-audit/SKILL.md +13 -7
- package/skills/bee/peaks-prd/SKILL.md +13 -7
- package/skills/bee/peaks-qa/SKILL.md +13 -7
- package/skills/bee/peaks-rd/SKILL.md +13 -7
- package/skills/bee/peaks-reviewer/SKILL.md +13 -7
- package/skills/bee/peaks-sc/SKILL.md +13 -7
- package/skills/bee/peaks-security-audit/SKILL.md +13 -7
- package/skills/bee/peaks-txt/SKILL.md +13 -7
- package/skills/bee/peaks-ui/SKILL.md +13 -7
- package/skills/peaks-audit/SKILL.md +13 -7
- package/skills/peaks-code/SKILL.md +13 -7
- package/skills/peaks-content/SKILL.md +13 -7
- package/skills/peaks-doctor/SKILL.md +13 -7
- package/skills/peaks-final-review/SKILL.md +13 -7
- package/skills/peaks-ide/SKILL.md +13 -7
- package/skills/peaks-issue-fix-orchestrator/SKILL.md +13 -7
- package/skills/peaks-resume/SKILL.md +13 -7
- package/skills/peaks-slice-decompose/SKILL.md +13 -7
- package/skills/peaks-solo/SKILL.md +13 -7
- package/skills/peaks-sop/SKILL.md +13 -7
- package/skills/peaks-status/SKILL.md +13 -7
- package/skills/peaks-test/SKILL.md +13 -7
package/CHANGELOG.md
CHANGED
|
@@ -1,5 +1,19 @@
|
|
|
1
1
|
# Changelog
|
|
2
2
|
|
|
3
|
+
## 4.0.41 — 2026-09-11 (闸门指导不再鼓励"每次编辑前念一遍")
|
|
4
|
+
|
|
5
|
+
**Highlights**:
|
|
6
|
+
|
|
7
|
+
1. **4.0.40 的措辞修好了"误读",但没修"频率" —— 这一版修频率。** 用户在 Mac 上继续工作时反馈:闸门仍然频繁打断。查下去发现**真正被烧掉的时间不是拒绝本身,而是一个仪式** —— 被告诉"预期每个文件被拦一次"之后,模型很自然地选择**预防**:在每次编辑前先把那四条事实念一遍,好让拒绝根本不出现。用户日志里那些重复的 `[Fact-Force Gate] 1..4` 块**是模型自己的话,不是闸门的输出**(闸门自称 `Fact-Forcing`,列的问题也不同)。每次编辑多一次「思考 10 秒 + 复述」,累起来就是他看到的 5 分钟。
|
|
8
|
+
|
|
9
|
+
- 现在指导明确要求:**不要预防性复述,闸门问的时候再答。** 拒绝是减速带,不是飞行前检查单。
|
|
10
|
+
|
|
11
|
+
2. **顺带记录了一条会读成"我退步了"的闸门行为。** 闲置约 30 分钟后,钩子会**删掉自己的状态文件**(`SESSION_TIMEOUT_MS`),清空"已通过"清单 —— 于是**之前已经放行的文件会再次被拦**。不知情的话,这读起来像模型又犯了错。指导里现在写明了:那是闸门在重置,不是你在退步。
|
|
12
|
+
|
|
13
|
+
3. **同时记下三个可用的旋钮**(闸门自带,非 peaks 引入):`GATEGUARD_EXEMPT_GLOBS`(逗号分隔 glob,加进去的路径直接豁免 —— peaks 写入时**取并集**,不会覆盖你写的)、`GATEGUARD_FACT_FORCE_FULL_DENIALS`(拒绝预算)、`GATEGUARD_DISABLED=1`(整个关掉)。**测试目录这类"没什么事实可建立"的路径,加一条 glob 就不再拦**。
|
|
14
|
+
|
|
15
|
+
**验证**:三个版本常量一致(4.0.41);`tsc -p tsconfig.build.json` exit 0;宽 `tsconfig.json` 保持 **142** 基线;`tests/unit` **188 files / 1794 passed / 3 skipped / 0 failed**。
|
|
16
|
+
|
|
3
17
|
## 4.0.40 — 2026-09-11 (两条被 4.0.39 暴露出来的既有缺陷)
|
|
4
18
|
|
|
5
19
|
**Highlights**:
|
package/README-en.md
CHANGED
|
@@ -140,7 +140,7 @@ Every lane opens with **one slash command**.
|
|
|
140
140
|
|
|
141
141
|
| | |
|
|
142
142
|
| --- | --- |
|
|
143
|
-
| **Latest** | [](https://www.npmjs.com/package/peaks-loop) — 4.0.
|
|
143
|
+
| **Latest** | [](https://www.npmjs.com/package/peaks-loop) — 4.0.41 (2026-09-11) |
|
|
144
144
|
| **Domains** | Code (`peaks-code`) · Content (`peaks-content`) · Project health (`peaks-doctor`) · Issue sweep (`peaks-issue-fix-orchestrator`) · Custom SOP (`peaks-sop`) · Cross-domain primitives (`peaks-solo` dispatcher · `peaks-resume` · `peaks-status` · `peaks-test` · `peaks-slice-decompose`) |
|
|
145
145
|
| **Sediment pool** | `~/.peaks/` local pool · twice-clean runs auto-promote to a bee · broken runs come back for you to redefine · the bee grows with your taste |
|
|
146
146
|
| **Test suite** | 285+ cases · 4 packages (peaks-loop / peaks-loop-mut / peaks-loop-shared-channel / peaks-loop-shared) · **0 timeouts** · 14 BDD caller-binding edge cases |
|
package/README.md
CHANGED
|
@@ -140,7 +140,7 @@ npm i -g peaks-loop
|
|
|
140
140
|
|
|
141
141
|
| | |
|
|
142
142
|
| --- | --- |
|
|
143
|
-
| **最新版本** | [](https://www.npmjs.com/package/peaks-loop) — 4.0.
|
|
143
|
+
| **最新版本** | [](https://www.npmjs.com/package/peaks-loop) — 4.0.41(2026-09-11) |
|
|
144
144
|
| **覆盖域** | 代码(`peaks-code`) · 内容(`peaks-content`) · 项目健康(`peaks-doctor`) · 批量修 issue(`peaks-issue-fix-orchestrator`) · 自定义 SOP(`peaks-sop`) · 通用原语(`peaks-solo` 分诊 / `peaks-resume` 续 / `peaks-status` 看 / `peaks-test` 测 / `peaks-slice-decompose` 切片) |
|
|
145
145
|
| **沉淀池** | `~/.peaks/` 本地池 · 跑两次自动晋升成 bee · 跑翻车让你重定义 · bee 跟着你的口味长 |
|
|
146
146
|
| **测试套件** | 1096 cases · 4 packages (peaks-loop 1015 / runtime 39 / mut 22 / shared-channel 20) · **CI 首次全绿**(ubuntu + windows) · 14 BDD caller-binding coverage |
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "peaks-loop",
|
|
3
|
-
"version": "4.0.
|
|
3
|
+
"version": "4.0.41",
|
|
4
4
|
"description": "Loop Engineering CLI — workflow primitive / loop guards / evaluators / slice orchestration",
|
|
5
5
|
"author": "SquabbyZ",
|
|
6
6
|
"keywords": [
|
|
@@ -101,10 +101,10 @@
|
|
|
101
101
|
"fzf": "^0.5.2",
|
|
102
102
|
"yaml": "^2.9.0",
|
|
103
103
|
"zod": "^4.4.3",
|
|
104
|
-
"peaks-loop-internal-runtime": "0.0.
|
|
105
|
-
"peaks-loop-shared": "0.0.
|
|
106
|
-
"peaks-loop-
|
|
107
|
-
"peaks-loop-
|
|
104
|
+
"peaks-loop-internal-runtime": "0.0.26",
|
|
105
|
+
"peaks-loop-shared-channel": "0.0.43",
|
|
106
|
+
"peaks-loop-mut": "0.1.39",
|
|
107
|
+
"peaks-loop-shared": "0.0.75"
|
|
108
108
|
},
|
|
109
109
|
"devDependencies": {
|
|
110
110
|
"@changesets/cli": "2.31.1",
|
|
@@ -31,13 +31,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
31
31
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
32
32
|
config); `.peaks/**` writes are exempt.
|
|
33
33
|
|
|
34
|
-
**Expect one denial per file
|
|
35
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
36
|
-
prevent it
|
|
37
|
-
|
|
38
|
-
|
|
39
|
-
|
|
40
|
-
|
|
34
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
35
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
36
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
37
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
38
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
39
|
+
denial actually arrives.
|
|
40
|
+
|
|
41
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
42
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
43
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
44
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
45
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
46
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
41
47
|
<!-- /peaks:loop-hygiene -->
|
|
42
48
|
|
|
43
49
|
# peaks-perf-audit
|
|
@@ -27,13 +27,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
27
27
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
28
28
|
config); `.peaks/**` writes are exempt.
|
|
29
29
|
|
|
30
|
-
**Expect one denial per file
|
|
31
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
32
|
-
prevent it
|
|
33
|
-
|
|
34
|
-
|
|
35
|
-
|
|
36
|
-
|
|
30
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
31
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
32
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
33
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
34
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
35
|
+
denial actually arrives.
|
|
36
|
+
|
|
37
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
38
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
39
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
40
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
41
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
42
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
37
43
|
<!-- /peaks:loop-hygiene -->
|
|
38
44
|
---
|
|
39
45
|
|
|
@@ -27,13 +27,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
27
27
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
28
28
|
config); `.peaks/**` writes are exempt.
|
|
29
29
|
|
|
30
|
-
**Expect one denial per file
|
|
31
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
32
|
-
prevent it
|
|
33
|
-
|
|
34
|
-
|
|
35
|
-
|
|
36
|
-
|
|
30
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
31
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
32
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
33
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
34
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
35
|
+
denial actually arrives.
|
|
36
|
+
|
|
37
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
38
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
39
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
40
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
41
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
42
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
37
43
|
<!-- /peaks:loop-hygiene -->
|
|
38
44
|
---
|
|
39
45
|
|
|
@@ -28,13 +28,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
28
28
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
29
29
|
config); `.peaks/**` writes are exempt.
|
|
30
30
|
|
|
31
|
-
**Expect one denial per file
|
|
32
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
33
|
-
prevent it
|
|
34
|
-
|
|
35
|
-
|
|
36
|
-
|
|
37
|
-
|
|
31
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
32
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
33
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
34
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
35
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
36
|
+
denial actually arrives.
|
|
37
|
+
|
|
38
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
39
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
40
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
41
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
42
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
43
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
38
44
|
<!-- /peaks:loop-hygiene -->
|
|
39
45
|
---
|
|
40
46
|
|
|
@@ -23,13 +23,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
23
23
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
24
24
|
config); `.peaks/**` writes are exempt.
|
|
25
25
|
|
|
26
|
-
**Expect one denial per file
|
|
27
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
28
|
-
prevent it
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
32
|
-
|
|
26
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
27
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
28
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
29
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
30
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
31
|
+
denial actually arrives.
|
|
32
|
+
|
|
33
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
34
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
35
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
36
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
37
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
38
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
33
39
|
<!-- /peaks:loop-hygiene -->
|
|
34
40
|
|
|
35
41
|
## Single-scope-axis naming convention
|
|
@@ -27,13 +27,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
27
27
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
28
28
|
config); `.peaks/**` writes are exempt.
|
|
29
29
|
|
|
30
|
-
**Expect one denial per file
|
|
31
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
32
|
-
prevent it
|
|
33
|
-
|
|
34
|
-
|
|
35
|
-
|
|
36
|
-
|
|
30
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
31
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
32
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
33
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
34
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
35
|
+
denial actually arrives.
|
|
36
|
+
|
|
37
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
38
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
39
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
40
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
41
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
42
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
37
43
|
<!-- /peaks:loop-hygiene -->
|
|
38
44
|
---
|
|
39
45
|
|
|
@@ -31,13 +31,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
31
31
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
32
32
|
config); `.peaks/**` writes are exempt.
|
|
33
33
|
|
|
34
|
-
**Expect one denial per file
|
|
35
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
36
|
-
prevent it
|
|
37
|
-
|
|
38
|
-
|
|
39
|
-
|
|
40
|
-
|
|
34
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
35
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
36
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
37
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
38
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
39
|
+
denial actually arrives.
|
|
40
|
+
|
|
41
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
42
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
43
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
44
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
45
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
46
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
41
47
|
<!-- /peaks:loop-hygiene -->
|
|
42
48
|
|
|
43
49
|
# peaks-security-audit
|
|
@@ -27,13 +27,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
27
27
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
28
28
|
config); `.peaks/**` writes are exempt.
|
|
29
29
|
|
|
30
|
-
**Expect one denial per file
|
|
31
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
32
|
-
prevent it
|
|
33
|
-
|
|
34
|
-
|
|
35
|
-
|
|
36
|
-
|
|
30
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
31
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
32
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
33
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
34
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
35
|
+
denial actually arrives.
|
|
36
|
+
|
|
37
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
38
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
39
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
40
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
41
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
42
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
37
43
|
<!-- /peaks:loop-hygiene -->
|
|
38
44
|
---
|
|
39
45
|
|
|
@@ -27,13 +27,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
27
27
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
28
28
|
config); `.peaks/**` writes are exempt.
|
|
29
29
|
|
|
30
|
-
**Expect one denial per file
|
|
31
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
32
|
-
prevent it
|
|
33
|
-
|
|
34
|
-
|
|
35
|
-
|
|
36
|
-
|
|
30
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
31
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
32
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
33
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
34
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
35
|
+
denial actually arrives.
|
|
36
|
+
|
|
37
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
38
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
39
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
40
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
41
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
42
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
37
43
|
<!-- /peaks:loop-hygiene -->
|
|
38
44
|
---
|
|
39
45
|
|
|
@@ -22,13 +22,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
22
22
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
23
23
|
config); `.peaks/**` writes are exempt.
|
|
24
24
|
|
|
25
|
-
**Expect one denial per file
|
|
26
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
-
prevent it
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
25
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
26
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
28
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
29
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
30
|
+
denial actually arrives.
|
|
31
|
+
|
|
32
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
33
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
34
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
35
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
36
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
37
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
32
38
|
<!-- /peaks:loop-hygiene -->
|
|
33
39
|
|
|
34
40
|
## Single-scope-axis naming convention
|
|
@@ -22,13 +22,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
22
22
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
23
23
|
config); `.peaks/**` writes are exempt.
|
|
24
24
|
|
|
25
|
-
**Expect one denial per file
|
|
26
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
-
prevent it
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
25
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
26
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
28
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
29
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
30
|
+
denial actually arrives.
|
|
31
|
+
|
|
32
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
33
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
34
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
35
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
36
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
37
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
32
38
|
<!-- /peaks:loop-hygiene -->
|
|
33
39
|
|
|
34
40
|
> **Detached sub-agent mode (Phase A, slice 2026-08-10).** When the orchestrator requires true parallelism with isolated context windows and survives orchestrator session exit, dispatch sub-agents with `--mode detached --vendor <claude|codex|copilot>`. The CLI spawns a real OS process via `ProcessSupervisor` (Windows `DETACHED_PROCESS` + `CREATE_NEW_PROCESS_GROUP`, POSIX `setsid` + `nohup`); the child vendor LLM receives a 5–8KB minimum prompt slice (no orchestrator session history) and self-compacts at 0.85 / 0.95 against the vendor window via the `<peaks-auto-compact>` marker (G8 — unlimited spend authorized). Orchestrator MUST emit one line of prose before every detached dispatch: `⏳ Spawning detached sub-agent via <vendor>: rid=<rid> (ETA ~60s)`. Status is read from `.peaks/_runtime/<sid>/detached/<rid>/status.json`; `LifecycleOwner` enforces 100% cleanup of `pid` / `log.txt` / `status.json` / `owner-session` on every exit path. `--no-throttle --max-concurrent <N>` bypasses `ResourceBudgetGuard` (user accepts risk; default max=8). See `references/sub-agent-dispatch.md` §"Detached Mode" for the full contract. Default mode remains `in-process` for backward compat (existing 106+ dispatch tests untouched).
|
|
@@ -22,13 +22,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
22
22
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
23
23
|
config); `.peaks/**` writes are exempt.
|
|
24
24
|
|
|
25
|
-
**Expect one denial per file
|
|
26
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
-
prevent it
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
25
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
26
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
28
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
29
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
30
|
+
denial actually arrives.
|
|
31
|
+
|
|
32
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
33
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
34
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
35
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
36
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
37
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
32
38
|
<!-- /peaks:loop-hygiene -->
|
|
33
39
|
|
|
34
40
|
# peaks-content
|
|
@@ -23,13 +23,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
23
23
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
24
24
|
config); `.peaks/**` writes are exempt.
|
|
25
25
|
|
|
26
|
-
**Expect one denial per file
|
|
27
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
28
|
-
prevent it
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
32
|
-
|
|
26
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
27
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
28
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
29
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
30
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
31
|
+
denial actually arrives.
|
|
32
|
+
|
|
33
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
34
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
35
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
36
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
37
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
38
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
33
39
|
<!-- /peaks:loop-hygiene -->
|
|
34
40
|
---
|
|
35
41
|
|
|
@@ -22,13 +22,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
22
22
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
23
23
|
config); `.peaks/**` writes are exempt.
|
|
24
24
|
|
|
25
|
-
**Expect one denial per file
|
|
26
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
-
prevent it
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
25
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
26
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
28
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
29
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
30
|
+
denial actually arrives.
|
|
31
|
+
|
|
32
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
33
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
34
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
35
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
36
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
37
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
32
38
|
<!-- /peaks:loop-hygiene -->
|
|
33
39
|
|
|
34
40
|
## Single-scope-axis naming convention
|
|
@@ -23,13 +23,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
23
23
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
24
24
|
config); `.peaks/**` writes are exempt.
|
|
25
25
|
|
|
26
|
-
**Expect one denial per file
|
|
27
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
28
|
-
prevent it
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
32
|
-
|
|
26
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
27
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
28
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
29
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
30
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
31
|
+
denial actually arrives.
|
|
32
|
+
|
|
33
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
34
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
35
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
36
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
37
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
38
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
33
39
|
<!-- /peaks:loop-hygiene -->
|
|
34
40
|
---
|
|
35
41
|
|
|
@@ -22,13 +22,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
22
22
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
23
23
|
config); `.peaks/**` writes are exempt.
|
|
24
24
|
|
|
25
|
-
**Expect one denial per file
|
|
26
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
-
prevent it
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
25
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
26
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
28
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
29
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
30
|
+
denial actually arrives.
|
|
31
|
+
|
|
32
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
33
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
34
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
35
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
36
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
37
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
32
38
|
<!-- /peaks:loop-hygiene -->
|
|
33
39
|
|
|
34
40
|
# peaks-issue-fix-orchestrator
|
|
@@ -22,13 +22,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
22
22
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
23
23
|
config); `.peaks/**` writes are exempt.
|
|
24
24
|
|
|
25
|
-
**Expect one denial per file
|
|
26
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
-
prevent it
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
25
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
26
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
28
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
29
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
30
|
+
denial actually arrives.
|
|
31
|
+
|
|
32
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
33
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
34
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
35
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
36
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
37
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
32
38
|
<!-- /peaks:loop-hygiene -->
|
|
33
39
|
|
|
34
40
|
# Peaks-Loop Code Resume (wrapper)
|
|
@@ -22,13 +22,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
22
22
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
23
23
|
config); `.peaks/**` writes are exempt.
|
|
24
24
|
|
|
25
|
-
**Expect one denial per file
|
|
26
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
-
prevent it
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
25
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
26
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
28
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
29
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
30
|
+
denial actually arrives.
|
|
31
|
+
|
|
32
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
33
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
34
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
35
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
36
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
37
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
32
38
|
<!-- /peaks:loop-hygiene -->
|
|
33
39
|
|
|
34
40
|
## Single-scope-axis naming convention
|
|
@@ -33,13 +33,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
33
33
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
34
34
|
config); `.peaks/**` writes are exempt.
|
|
35
35
|
|
|
36
|
-
**Expect one denial per file
|
|
37
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
38
|
-
prevent it
|
|
39
|
-
|
|
40
|
-
|
|
41
|
-
|
|
42
|
-
|
|
36
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
37
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
38
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
39
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
40
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
41
|
+
denial actually arrives.
|
|
42
|
+
|
|
43
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
44
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
45
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
46
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
47
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
48
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
43
49
|
<!-- /peaks:loop-hygiene -->
|
|
44
50
|
|
|
45
51
|
# peaks-solo — Dispatcher (分诊员)
|
|
@@ -22,13 +22,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
22
22
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
23
23
|
config); `.peaks/**` writes are exempt.
|
|
24
24
|
|
|
25
|
-
**Expect one denial per file
|
|
26
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
-
prevent it
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
25
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
26
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
28
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
29
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
30
|
+
denial actually arrives.
|
|
31
|
+
|
|
32
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
33
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
34
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
35
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
36
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
37
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
32
38
|
<!-- /peaks:loop-hygiene -->
|
|
33
39
|
|
|
34
40
|
# Peaks-Loop SOP Authoring
|
|
@@ -22,13 +22,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
22
22
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
23
23
|
config); `.peaks/**` writes are exempt.
|
|
24
24
|
|
|
25
|
-
**Expect one denial per file
|
|
26
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
-
prevent it
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
25
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
26
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
28
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
29
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
30
|
+
denial actually arrives.
|
|
31
|
+
|
|
32
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
33
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
34
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
35
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
36
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
37
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
32
38
|
<!-- /peaks:loop-hygiene -->
|
|
33
39
|
|
|
34
40
|
# Peaks-Loop Code Status (wrapper)
|
|
@@ -22,13 +22,19 @@ mode** — standard *and* 24h. The mode moves the threshold, never the obligatio
|
|
|
22
22
|
it — the normal way to work here — for every path outside `.peaks/**` (source, tests, docs,
|
|
23
23
|
config); `.peaks/**` writes are exempt.
|
|
24
24
|
|
|
25
|
-
**Expect one denial per file
|
|
26
|
-
denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
-
prevent it
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
25
|
+
**Expect one denial per file anyway — and do NOT pre-empt it.** A `PreToolUse` gate (ECC's
|
|
26
|
+
"Fact-Forcing Gate") denies the FIRST edit a session makes to any given path, once, by
|
|
27
|
+
design. Reading does NOT prevent it: the gate keys on the path's first touch, not on whether
|
|
28
|
+
you read it. **Do not recite its four questions before every edit** — it asks when it wants
|
|
29
|
+
them, and reciting unprompted burns a round-trip per file for nothing. Answer only when a
|
|
30
|
+
denial actually arrives.
|
|
31
|
+
|
|
32
|
+
**When one does arrive:** a denial is not a failure and the tool is not broken — your edit
|
|
33
|
+
was NOT applied. State the facts it asks for (importers, affected API, data schemas if any,
|
|
34
|
+
the user's verbatim instruction), then retry the SAME operation. The retry is allowed. Do not
|
|
35
|
+
switch tools, do not give up, do not retry blindly. One more thing worth knowing: an idle gap
|
|
36
|
+
of ~30 minutes clears the gate's "already passed" list, so a file you cleared earlier can be
|
|
37
|
+
denied again after a long pause. That is the gate resetting, not you regressing.
|
|
32
38
|
<!-- /peaks:loop-hygiene -->
|
|
33
39
|
|
|
34
40
|
# Peaks-Loop Code Test (wrapper)
|