@adamchanadam/agent-handoff-kit 0.3.41 → 0.3.43
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.
- package/LICENSE +21 -21
- package/README.en.md +81 -0
- package/README.md +38 -36
- package/bin/agent-handoff-kit.mjs +4664 -2544
- package/bin/installed-file-contract.mjs +74 -45
- package/bin/migration-baselines/official-origin-catalog.json +10039 -9828
- package/bin/official-origin-catalog.mjs +248 -192
- package/bin/prompt-mirror-core.mjs +84 -84
- package/bin/semantic-candidate.mjs +594 -0
- package/bin/upgrade-inventory.mjs +587 -0
- package/bin/user-rules-router.mjs +461 -0
- package/package.json +1 -13
- package/packs/agent-governance.md +103 -103
- package/packs/closeout.md +74 -72
- package/packs/coding.md +28 -28
- package/packs/communication.md +32 -30
- package/packs/integrations.md +251 -251
- package/packs/knowledge.md +5 -5
- package/packs/onboarding.md +214 -214
- package/packs/release.md +28 -28
- package/packs/research.md +30 -30
- package/packs/safety.md +29 -29
- package/packs/writing.md +36 -36
- package/runtime-core/AGENTS.core.md +115 -112
- package/runtime-core/CLAUDE.md +13 -13
- package/runtime-core/DOC_SYNC_REGISTRY.md +24 -24
- package/runtime-core/GEMINI.md +17 -17
- package/runtime-core/PROJECT_DECISIONS.md +43 -43
- package/runtime-core/PROJECT_INDEX.md +73 -73
- package/runtime-core/RULE_PACKS.md +14 -14
- package/runtime-core/SESSION_HANDOFF.md +42 -36
- package/runtime-core/SESSION_LOG.md +35 -35
- package/runtime-core/START_NEXT_SESSION_PROMPT.txt +5 -5
- package/runtime-core/USER_RULES.md +30 -0
- package/runtime-core/USER_RULES_ENTRY.md +7 -0
package/LICENSE
CHANGED
|
@@ -1,21 +1,21 @@
|
|
|
1
|
-
MIT License
|
|
2
|
-
|
|
3
|
-
Copyright (c) 2026 Prompt Templates
|
|
4
|
-
|
|
5
|
-
Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
6
|
-
of this software and associated documentation files (the "Software"), to deal
|
|
7
|
-
in the Software without restriction, including without limitation the rights
|
|
8
|
-
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
9
|
-
copies of the Software, and to permit persons to whom the Software is
|
|
10
|
-
furnished to do so, subject to the following conditions:
|
|
11
|
-
|
|
12
|
-
The above copyright notice and this permission notice shall be included in all
|
|
13
|
-
copies or substantial portions of the Software.
|
|
14
|
-
|
|
15
|
-
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
16
|
-
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
17
|
-
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
18
|
-
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
19
|
-
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
20
|
-
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
21
|
-
SOFTWARE.
|
|
1
|
+
MIT License
|
|
2
|
+
|
|
3
|
+
Copyright (c) 2026 Prompt Templates
|
|
4
|
+
|
|
5
|
+
Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
6
|
+
of this software and associated documentation files (the "Software"), to deal
|
|
7
|
+
in the Software without restriction, including without limitation the rights
|
|
8
|
+
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
9
|
+
copies of the Software, and to permit persons to whom the Software is
|
|
10
|
+
furnished to do so, subject to the following conditions:
|
|
11
|
+
|
|
12
|
+
The above copyright notice and this permission notice shall be included in all
|
|
13
|
+
copies or substantial portions of the Software.
|
|
14
|
+
|
|
15
|
+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
16
|
+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
17
|
+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
18
|
+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
19
|
+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
20
|
+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
21
|
+
SOFTWARE.
|
package/README.en.md
ADDED
|
@@ -0,0 +1,81 @@
|
|
|
1
|
+
# Agent Handoff Kit
|
|
2
|
+
|
|
3
|
+
[繁體中文](README.md) · [Getting started](https://adamchanadam.github.io/agent-handoff-kit/agent-handoff-kit-intro.en.html) · [Practical guide](https://adamchanadam.github.io/agent-handoff-kit/agent-handoff-kit-guide.en.html) · [AI install page](https://adamchanadam.github.io/agent-handoff-kit/agent-handoff-kit-ai-install.en.html)
|
|
4
|
+
|
|
5
|
+
Current candidate: `v0.3.43` (not published). The currently published npm release is `v0.3.42`.
|
|
6
|
+
|
|
7
|
+
Agent Handoff Kit is a small continuity layer for local AI agents. It gives the next AI conversation an honest, readable project handoff: what happened, what matters next, what must not be broken, and which project files are authoritative.
|
|
8
|
+
|
|
9
|
+
It is for AI tools that can read and write your local project folder. It is not a prompt for ordinary web-chat AI that cannot access that folder.
|
|
10
|
+
|
|
11
|
+
## Start in three steps
|
|
12
|
+
|
|
13
|
+
1. Open an AI agent in the folder you want to work in and send it this message:
|
|
14
|
+
|
|
15
|
+
```text
|
|
16
|
+
Read https://adamchanadam.github.io/agent-handoff-kit/agent-handoff-kit-ai-install.en.html and install or upgrade Agent Handoff Kit in this folder.
|
|
17
|
+
```
|
|
18
|
+
|
|
19
|
+
2. After installation, say `Start Agent Handoff` or `開工` to that AI.
|
|
20
|
+
3. When you are genuinely ending the session, say `wrap up` or `收工`.
|
|
21
|
+
|
|
22
|
+
You do not need to decide whether this is an install, upgrade, health check, or conflict. The AI must first show the folder it sees and ask for confirmation before it writes. Existing AI files such as `AGENTS.md`, `CLAUDE.md`, and `GEMINI.md` are inspected rather than silently overwritten.
|
|
23
|
+
|
|
24
|
+
## Choose an entry point
|
|
25
|
+
|
|
26
|
+
| Page | What it is for |
|
|
27
|
+
| --- | --- |
|
|
28
|
+
| [This README](README.en.md) | Product purpose, safe use, and boundaries. |
|
|
29
|
+
| [60-second introduction](https://adamchanadam.github.io/agent-handoff-kit/agent-handoff-kit-intro.en.html) | A non-technical overview. |
|
|
30
|
+
| [Practical guide](https://adamchanadam.github.io/agent-handoff-kit/agent-handoff-kit-guide.en.html) | Three everyday scenarios from installation to handoff. |
|
|
31
|
+
| [AI install page](https://adamchanadam.github.io/agent-handoff-kit/agent-handoff-kit-ai-install.en.html) | The precise instructions an AI uses to install, upgrade, or stop safely. |
|
|
32
|
+
|
|
33
|
+
## What problem it solves
|
|
34
|
+
|
|
35
|
+
Long-running AI work usually fails at the boundaries between conversations:
|
|
36
|
+
|
|
37
|
+
| Problem | What the kit does |
|
|
38
|
+
| --- | --- |
|
|
39
|
+
| The next AI does not know the current state | Keeps a current handoff with objective, risks, validation, and next action. |
|
|
40
|
+
| New documents become disconnected | Routes important documents through the project index and rule packs. |
|
|
41
|
+
| A session ends without a usable handoff | A real closeout records the result and refreshes the next-session prompt. |
|
|
42
|
+
| An upgrade overwrites local material | Only exact, artifact-identified Kit content is replaceable; non-exact content is preserved or causes a safe stop. |
|
|
43
|
+
| A health check says success but gives contradictory advice | Doctor reads the same verified accepted state as the transaction, report, and success result. |
|
|
44
|
+
|
|
45
|
+
## What an upgrade promises
|
|
46
|
+
|
|
47
|
+
An upgrade must not trade safety for convenience. It must preserve user and user-AI content, update only exact Kit-managed bytes, keep effective rules reachable through the normal runtime entry, and recover to a complete old or complete new state if interrupted.
|
|
48
|
+
|
|
49
|
+
The published upgrade path is checked with a previous-version artifact, direct user rules in `AGENTS.md`, fresh formal user rules, repeated upgrades, bytes drift, and recovery. If ownership cannot be proved exactly, the tool must not guess from a title, language, location, or formatting.
|
|
50
|
+
|
|
51
|
+
## Everyday use
|
|
52
|
+
|
|
53
|
+
Use ordinary language. Examples:
|
|
54
|
+
|
|
55
|
+
- `Start Agent Handoff`
|
|
56
|
+
- `Help me review this proposal and keep the project handoff current.`
|
|
57
|
+
- `Please connect docs/research-notes.md to Agent Handoff Kit.`
|
|
58
|
+
- `wrap up`
|
|
59
|
+
|
|
60
|
+
A bare `Start Agent Handoff` or `開工` restores the minimum current state, shows the startup card and recommended next action, then stops. It does not authorize the AI to begin research, planning, QA, packaging, or file edits on its own. To continue work immediately, say so in the same message, for example: `Start Agent Handoff, continue the current task.`
|
|
61
|
+
|
|
62
|
+
## Technical boundary
|
|
63
|
+
|
|
64
|
+
The npm package contains only the runtime files needed by a project: the CLI, runtime core, and rule packs. Public source documentation and release QA live in the source repository and are deliberately not installed into user projects.
|
|
65
|
+
|
|
66
|
+
For manual terminal use, the canonical commands are:
|
|
67
|
+
|
|
68
|
+
```text
|
|
69
|
+
npx --yes @adamchanadam/agent-handoff-kit@latest init
|
|
70
|
+
npx --yes @adamchanadam/agent-handoff-kit@latest upgrade --dry-run
|
|
71
|
+
npx --yes @adamchanadam/agent-handoff-kit@latest doctor
|
|
72
|
+
```
|
|
73
|
+
|
|
74
|
+
`upgrade --dry-run` is only a preview. It is not a completed upgrade.
|
|
75
|
+
|
|
76
|
+
## Links
|
|
77
|
+
|
|
78
|
+
- [GitHub repository](https://github.com/Adamchanadam/agent-handoff-kit)
|
|
79
|
+
- [npm package](https://www.npmjs.com/package/@adamchanadam/agent-handoff-kit)
|
|
80
|
+
- [Traditional Chinese README](README.md)
|
|
81
|
+
- [License](LICENSE)
|
package/README.md
CHANGED
|
@@ -1,6 +1,8 @@
|
|
|
1
|
-
# Agent Handoff Kit
|
|
2
|
-
|
|
3
|
-
|
|
1
|
+
# Agent Handoff Kit
|
|
2
|
+
|
|
3
|
+
English: [README.en.md](README.en.md) · [Getting started](https://adamchanadam.github.io/agent-handoff-kit/agent-handoff-kit-intro.en.html) · [Practical guide](https://adamchanadam.github.io/agent-handoff-kit/agent-handoff-kit-guide.en.html) · [AI install page](https://adamchanadam.github.io/agent-handoff-kit/agent-handoff-kit-ai-install.en.html)
|
|
4
|
+
|
|
5
|
+
目前候選版本:`v0.3.43`(尚未發佈);正式可用版本仍是 `v0.3.42`。
|
|
4
6
|
|
|
5
7
|
<p align="center">
|
|
6
8
|
<img src="https://raw.githubusercontent.com/Adamchanadam/agent-handoff-kit/main/images/agent-handoff-kit-promo-30s.gif" alt="Agent Handoff Kit 功能簡介動畫" width="720">
|
|
@@ -8,11 +10,11 @@
|
|
|
8
10
|
|
|
9
11
|
Agent Handoff Kit 是 **AI 對話之間的接力棒**。
|
|
10
12
|
|
|
11
|
-
它只處理一件狹窄但重要的事:AI 跨對話失憶。每次開新對話,AI 往往不記得你上次做到哪裡,也認不出中途新建的文件、你引入的參考資料、哪份文件才是最新依據。這套工具把進度、下一步、風險、檔案登記與下次開工提示寫進固定文件,讓下一個 AI 工具能接得上上一棒。
|
|
13
|
+
它只處理一件狹窄但重要的事:AI 跨對話失憶。每次開新對話,AI 往往不記得你上次做到哪裡,也認不出中途新建的文件、你引入的參考資料、哪份文件才是最新依據。這套工具把進度、下一步、風險、檔案登記與下次開工提示寫進固定文件,讓下一個 AI 工具能接得上上一棒。
|
|
12
14
|
|
|
13
|
-
📌 使用時,你只需要說明目的;確認資料夾、判斷安裝或升級、執行指令和檢查結果,交給能讀寫本機資料夾的 AI 處理。
|
|
14
|
-
|
|
15
|
-
## 🚀 三步上手
|
|
15
|
+
📌 使用時,你只需要說明目的;確認資料夾、判斷安裝或升級、執行指令和檢查結果,交給能讀寫本機資料夾的 AI 處理。
|
|
16
|
+
|
|
17
|
+
## 🚀 三步上手
|
|
16
18
|
|
|
17
19
|
第一次用,不需要先讀完整 README,也不需要研究終端機指令。只做三件事:
|
|
18
20
|
|
|
@@ -39,27 +41,27 @@ Agent Handoff Kit 是 **AI 對話之間的接力棒**。
|
|
|
39
41
|
|
|
40
42
|
| 入口 | 用途 |
|
|
41
43
|
|---|---|
|
|
42
|
-
| `README.md` | 正式用途、安裝路徑與安全邊界。 |
|
|
44
|
+
| `README.md` | 正式用途、安裝路徑與安全邊界。 |
|
|
43
45
|
| [`agent-handoff-kit-intro.html`](https://adamchanadam.github.io/agent-handoff-kit/agent-handoff-kit-intro.html) | 非技術版 60 秒入門與宣傳動畫。 |
|
|
44
46
|
| [`agent-handoff-kit-guide.html`](https://adamchanadam.github.io/agent-handoff-kit/agent-handoff-kit-guide.html) | 三個實操情景,示範開工、工作、收工。 |
|
|
45
47
|
| [`agent-handoff-kit-ai-install.html`](https://adamchanadam.github.io/agent-handoff-kit/agent-handoff-kit-ai-install.html) | 給 AI 讀的安裝與升級指令頁。 |
|
|
46
48
|
|
|
47
|
-
這個公開 repo 保留使用、安裝、入門、GitHub Pages 與 npm 執行所需內容。日常使用時,你只需要從上面四個入口開始。
|
|
49
|
+
這個公開 repo 保留使用、安裝、入門、GitHub Pages 與 npm 執行所需內容。日常使用時,你只需要從上面四個入口開始。
|
|
48
50
|
|
|
49
51
|
## 🔎 它解決甚麼問題
|
|
50
52
|
|
|
51
|
-
用 AI 做長期項目,常見五個問題:
|
|
53
|
+
用 AI 做長期項目,常見五個問題:
|
|
52
54
|
|
|
53
55
|
| 問題 | Agent Handoff Kit 怎樣處理 |
|
|
54
56
|
|---|---|
|
|
55
57
|
| 新 AI 不知做到哪 | 用 `dev/SESSION_HANDOFF.md` 保存目前狀態、下一步、風險與驗收。 |
|
|
56
|
-
| 新建檔案、參考資料變孤兒 | 你可以叫 AI 把文件接入 Agent Handoff Kit,記低每份文件的用途、哪份作準、何時要同步。 |
|
|
57
|
-
| 不同 AI 工具入口不同 | 同時安裝 `AGENTS.md`、`CLAUDE.md`、`GEMINI.md`,全部指向同一套開工流程;Antigravity CLI 會讀工作資料夾內的 `AGENTS.md` 與 `GEMINI.md`。 |
|
|
58
|
-
| AI 可能亂改、亂刪或誤發佈 | 內置安全規則;高風險操作必須先講計劃,破壞性指令與未批准發佈一律禁止。 |
|
|
59
|
-
| 外部工具用完後殘留或亂清理 | 使用 MCP、browser、自動化工具、notebook 或 helper server 後,AI 會按 ownership 判斷:本任務資源可收口,不明或共享資源先回報證據並等你確認。 |
|
|
60
|
-
|
|
61
|
-
|
|
62
|
-
它不是聊天機器人,也不是開發框架。它比較像一本固定放在專案內的交接簿。
|
|
58
|
+
| 新建檔案、參考資料變孤兒 | 你可以叫 AI 把文件接入 Agent Handoff Kit,記低每份文件的用途、哪份作準、何時要同步。 |
|
|
59
|
+
| 不同 AI 工具入口不同 | 同時安裝 `AGENTS.md`、`CLAUDE.md`、`GEMINI.md`,全部指向同一套開工流程;Antigravity CLI 會讀工作資料夾內的 `AGENTS.md` 與 `GEMINI.md`。 |
|
|
60
|
+
| AI 可能亂改、亂刪或誤發佈 | 內置安全規則;高風險操作必須先講計劃,破壞性指令與未批准發佈一律禁止。 |
|
|
61
|
+
| 外部工具用完後殘留或亂清理 | 使用 MCP、browser、自動化工具、notebook 或 helper server 後,AI 會按 ownership 判斷:本任務資源可收口,不明或共享資源先回報證據並等你確認。 |
|
|
62
|
+
|
|
63
|
+
|
|
64
|
+
它不是聊天機器人,也不是開發框架。它比較像一本固定放在專案內的交接簿。
|
|
63
65
|
|
|
64
66
|
## 🧰 適用工具
|
|
65
67
|
|
|
@@ -85,9 +87,9 @@ Start Agent Handoff
|
|
|
85
87
|
Work in <你的專案資料夾>. Read AGENTS.md first, then Start Agent Handoff. Before changing anything, tell me the current state and your recommended next step.
|
|
86
88
|
```
|
|
87
89
|
|
|
88
|
-
第一次安裝只會令新手引導可用,不會強制啟動。你已清楚描述目標和現有資料時,AI 會直接開始第一個安全步驟;只有目標仍然含糊,或你明確要求教學時,才會載入新手引導。日常「開工」先讀 `dev/SESSION_HANDOFF.md`;同一資料夾內不會再重讀 `START_NEXT_SESSION_PROMPT.txt` 或 `dev/SESSION_LOG.md`。提示副本只供尚未指向專案資料夾的 AI 接入。
|
|
90
|
+
第一次安裝只會令新手引導可用,不會強制啟動。你已清楚描述目標和現有資料時,AI 會直接開始第一個安全步驟;只有目標仍然含糊,或你明確要求教學時,才會載入新手引導。日常「開工」先讀 `dev/SESSION_HANDOFF.md`;同一資料夾內不會再重讀 `START_NEXT_SESSION_PROMPT.txt` 或 `dev/SESSION_LOG.md`。提示副本只供尚未指向專案資料夾的 AI 接入。
|
|
89
91
|
|
|
90
|
-
「開工,繼續 <任務>」、「<項目> 開工」或 `Start Agent Handoff and continue <task>` 會直接接力。只有語句明顯可能指現實世界的開業、輪班或其他無關事件時,AI 才作一次簡短確認。
|
|
92
|
+
「開工,繼續 <任務>」、「<項目> 開工」或 `Start Agent Handoff and continue <task>` 會直接接力。只有語句明顯可能指現實世界的開業、輪班或其他無關事件時,AI 才作一次簡短確認。
|
|
91
93
|
|
|
92
94
|
然後用日常話描述你要完成的任務。AI 應先讀交接文件,說明目前狀態、下一步與風險,再開始工作。
|
|
93
95
|
|
|
@@ -145,9 +147,9 @@ AI 會處理檢查,不會把「檢查通過」誤當成已理解你的專案
|
|
|
145
147
|
|---|---|
|
|
146
148
|
| 寫程式 / 修錯誤 | 先讀項目索引、相關檔案,再改動和測試。 |
|
|
147
149
|
| 寫文章 / README / 社交帖文 | 先確認讀者、目的、語氣和發布位置。 |
|
|
148
|
-
| 查資料 / 比較工具 | 分清已驗證事實、來源摘要和 AI 的判斷。 |
|
|
150
|
+
| 查資料 / 比較工具 | 分清已驗證事實、來源摘要和 AI 的判斷。 |
|
|
149
151
|
| 刪檔 / Git / 發佈 / npm | 高風險操作必須先說明影響,並等你確認。 |
|
|
150
|
-
| Notion / Google Drive 等外部工具 | 先核對目前工具說明或官方文件;用後按資源歸屬安全收口;機密不寫入項目文件。 |
|
|
152
|
+
| Notion / Google Drive 等外部工具 | 先核對目前工具說明或官方文件;用後按資源歸屬安全收口;機密不寫入項目文件。 |
|
|
151
153
|
|
|
152
154
|
你只要用日常話說目的,例如「幫我改 README」、「幫我查這個工具是否適合」、「把這份文件接入 Agent Handoff Kit」。AI 會自己判斷要用哪些規則。
|
|
153
155
|
|
|
@@ -169,33 +171,33 @@ AI 應先判斷這條規則應放在哪裡:是一次性備忘、下次交接
|
|
|
169
171
|
| 結束本輪工作 | 「收工」 |
|
|
170
172
|
| 讓新文件不變成孤兒 | 「把這份文件接入 Agent Handoff Kit,讓下次 AI 知道何時要讀、何時要更新。」 |
|
|
171
173
|
| 掃描可能被遺漏的重要文件 | 「掃描未接入 Agent Handoff Kit 的重要文件。」 |
|
|
172
|
-
| 讓 AI 下次避免同類錯誤 | 「把今次錯誤整理成日後工作規則,讓下次 AI 知道要怎樣避免。」 |
|
|
173
|
-
| 讓 API / 工具用法下次仍生效 | 「以後都用這個 API 調用方式,之後開新對話也要沿用。」 |
|
|
174
|
-
| 使用外部工具,例如 Notion、Google Drive、GitHub | 「這個項目會用到這些外部工具,請記住哪些能直接使用,機密不要寫入項目文件。」 |
|
|
175
|
-
| 長任務用了 MCP、browser 或自動化工具 | 「收工時請顯示外部工具資源收口結果:已關閉哪些本任務資源,哪些因歸屬不明而保留並列證據。」 |
|
|
174
|
+
| 讓 AI 下次避免同類錯誤 | 「把今次錯誤整理成日後工作規則,讓下次 AI 知道要怎樣避免。」 |
|
|
175
|
+
| 讓 API / 工具用法下次仍生效 | 「以後都用這個 API 調用方式,之後開新對話也要沿用。」 |
|
|
176
|
+
| 使用外部工具,例如 Notion、Google Drive、GitHub | 「這個項目會用到這些外部工具,請記住哪些能直接使用,機密不要寫入項目文件。」 |
|
|
177
|
+
| 長任務用了 MCP、browser 或自動化工具 | 「收工時請顯示外部工具資源收口結果:已關閉哪些本任務資源,哪些因歸屬不明而保留並列證據。」 |
|
|
176
178
|
|
|
177
|
-
掃描重要文件時,AI 只會先列出可能需要接入的文件與原因,不會自動修改;是否接入、合併或退役由你確認。
|
|
178
|
-
|
|
179
|
-
需要長期保留的規則,AI 應寫入合適的項目文件,而不是只留在當次對話摘要。
|
|
179
|
+
掃描重要文件時,AI 只會先列出可能需要接入的文件與原因,不會自動修改;是否接入、合併或退役由你確認。
|
|
180
|
+
|
|
181
|
+
需要長期保留的規則,AI 應寫入合適的項目文件,而不是只留在當次對話摘要。
|
|
180
182
|
|
|
181
|
-
如涉及刪除、改名、合併權威文件、發佈、上傳或權限變更,AI 應先說明影響並等你確認。
|
|
183
|
+
如涉及刪除、改名、合併權威文件、發佈、上傳或權限變更,AI 應先說明影響並等你確認。
|
|
182
184
|
|
|
183
185
|
## 🛡️ 安全護欄
|
|
184
186
|
|
|
185
187
|
就算你不懂代碼,這套工具也會要求 AI 在高風險操作前停下來講清楚。
|
|
186
188
|
|
|
187
|
-
- 破壞性操作分界:`rm -rf`、`git reset --hard` 等指定破壞性命令及系統根路徑操作一律禁止;強制推送、分支或標籤刪除及歷史改寫,須另取明確授權並核對受影響 ref。
|
|
188
|
-
- 機密保護:`.env`、API key、token 不可印出、不可提交、不可上傳。
|
|
189
|
-
- 查證不猜:使用第三方服務、Connector、MCP、CLI、API 或 plugin API 前,先核對目前工具說明、官方文件或已驗證的本地操作說明;查不到就標示未核實。
|
|
190
|
-
- 外部工具安全收口:用完 MCP、browser、自動化工具、notebook 或 helper server 後,AI 只可自動關閉能證明屬於本任務的資源;不明、共享、使用者或其他 AI agent 可能擁有的程序與暫存資料,必須先回報並等你確認。
|
|
191
|
-
- 權限不足就停手:檔案被鎖或沒有權限時,輸出手動操作清單,不嘗試繞過。
|
|
189
|
+
- 破壞性操作分界:`rm -rf`、`git reset --hard` 等指定破壞性命令及系統根路徑操作一律禁止;強制推送、分支或標籤刪除及歷史改寫,須另取明確授權並核對受影響 ref。
|
|
190
|
+
- 機密保護:`.env`、API key、token 不可印出、不可提交、不可上傳。
|
|
191
|
+
- 查證不猜:使用第三方服務、Connector、MCP、CLI、API 或 plugin API 前,先核對目前工具說明、官方文件或已驗證的本地操作說明;查不到就標示未核實。
|
|
192
|
+
- 外部工具安全收口:用完 MCP、browser、自動化工具、notebook 或 helper server 後,AI 只可自動關閉能證明屬於本任務的資源;不明、共享、使用者或其他 AI agent 可能擁有的程序與暫存資料,必須先回報並等你確認。
|
|
193
|
+
- 權限不足就停手:檔案被鎖或沒有權限時,輸出手動操作清單,不嘗試繞過。
|
|
192
194
|
- 發佈需明確批准:建立版本標籤、GitHub Release、npm publish、部署或上傳,都不能因「準備好了」而自動執行。
|
|
193
195
|
|
|
194
196
|
## 🔗 可選配合:Adam-AI-Instructions
|
|
195
197
|
|
|
196
|
-
Agent Handoff Kit 可與 [Adam-AI-Instructions](https://github.com/prompt-templates/Adam-AI-Instructions) 配合使用。兩者分工互補,但在安全、機密、不可逆操作及發佈底線上刻意保留最低限度的共同防線;這不是父子真源,也不應合併成同一套規則:
|
|
198
|
+
Agent Handoff Kit 可與 [Adam-AI-Instructions](https://github.com/prompt-templates/Adam-AI-Instructions) 配合使用。兩者分工互補,但在安全、機密、不可逆操作及發佈底線上刻意保留最低限度的共同防線;這不是父子真源,也不應合併成同一套規則:
|
|
197
199
|
|
|
198
200
|
- **Adam-AI-Instructions** 負責 AI 在**單一對話**內的做事規矩:語氣、做事優先序、回覆骨架、計算紀律、用語紀律、安全護欄、輸出層分工。屬「AI 應該怎樣答你」的持久基準。
|
|
199
201
|
- **Agent Handoff Kit** 負責 AI 在**對話之間**的接力:當前狀態、下一步、檔案登記、收工同下次開工。屬「AI 在對話之間怎樣記住你的項目」的持久基準。
|
|
200
202
|
|
|
201
|
-
這是可選配合,不影響 Agent Handoff Kit 的安裝和日常使用。想使用時,到該倉庫選擇適合你 AI 工具的版本,貼入 AI 工具設定即可。
|
|
203
|
+
這是可選配合,不影響 Agent Handoff Kit 的安裝和日常使用。想使用時,到該倉庫選擇適合你 AI 工具的版本,貼入 AI 工具設定即可。
|