clitrigger 0.2.47 → 0.2.49
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/README.md +12 -16
- package/README_KR.md +12 -16
- package/dist/client/assets/{AnalyticsPanel-B6F3zRih.js → AnalyticsPanel-CKy5UZMe.js} +1 -1
- package/dist/client/assets/{FloatingComposerController-5A3GJtK4-BM6J3olW.js → FloatingComposerController-5A3GJtK4-BbmVg3RD.js} +1 -1
- package/dist/client/assets/{PlannerPageEditor-BiAaXcG8.js → PlannerPageEditor-DXOuwKvY.js} +3 -3
- package/dist/client/assets/{TaskGraph-HfQKz7kQ.js → TaskGraph-Ctknpw0b.js} +1 -1
- package/dist/client/assets/{VaultLayout-DUcZffQx.js → VaultLayout-C-tgc10Z.js} +1 -1
- package/dist/client/assets/{index-XgSru2ff.js → index-BHP8zs7Q.js} +179 -189
- package/dist/client/assets/index-BQW3mIMS.css +32 -0
- package/dist/client/assets/{style-BGfsFGOn.js → style-D864rZPo.js} +1 -1
- package/dist/client/assets/{with-selector-Cu88HUQY.js → with-selector-DYuPlN7R.js} +1 -1
- package/dist/client/index.html +2 -2
- package/dist/server/db/queries.d.ts +4 -2
- package/dist/server/db/queries.d.ts.map +1 -1
- package/dist/server/db/queries.js +7 -3
- package/dist/server/db/queries.js.map +1 -1
- package/dist/server/db/schema.d.ts.map +1 -1
- package/dist/server/db/schema.js +4 -0
- package/dist/server/db/schema.js.map +1 -1
- package/dist/server/routes/projects.d.ts.map +1 -1
- package/dist/server/routes/projects.js +7 -2
- package/dist/server/routes/projects.js.map +1 -1
- package/dist/server/services/auto-delegate.d.ts +16 -0
- package/dist/server/services/auto-delegate.d.ts.map +1 -0
- package/dist/server/services/auto-delegate.js +56 -0
- package/dist/server/services/auto-delegate.js.map +1 -0
- package/dist/server/services/orchestrator.d.ts.map +1 -1
- package/dist/server/services/orchestrator.js +13 -1
- package/dist/server/services/orchestrator.js.map +1 -1
- package/dist/server/websocket/events.d.ts +4 -0
- package/dist/server/websocket/events.d.ts.map +1 -1
- package/electron/main.cjs +41 -33
- package/package.json +1 -1
- package/dist/client/assets/index-PHcsJJs8.css +0 -32
package/README.md
CHANGED
|
@@ -6,9 +6,9 @@
|
|
|
6
6
|
<img alt="CLITrigger" src="https://raw.githubusercontent.com/HyperAITeam/CLITrigger/main/src/client/public/logo.svg" width="360">
|
|
7
7
|
</picture>
|
|
8
8
|
|
|
9
|
-
**
|
|
9
|
+
**An IDE for AI CLI Agents**
|
|
10
10
|
|
|
11
|
-
*
|
|
11
|
+
*Docs, plans, terminals, autonomous agents, and git — one workspace instead of five scattered tools.*
|
|
12
12
|
|
|
13
13
|
<p align="center">
|
|
14
14
|
<a href="https://github.com/HyperAITeam/CLITrigger/blob/main/README.md">English</a> ·
|
|
@@ -44,11 +44,11 @@ npm i -g clitrigger && clitrigger
|
|
|
44
44
|
|
|
45
45
|
> ### Docs → Plan → Terminal → Autonomous Tasks → Version Control. One pipeline.
|
|
46
46
|
>
|
|
47
|
-
>
|
|
47
|
+
> Developing with AI CLI agents (Claude Code, Codex, …) scatters the workflow across disconnected tools: requirements in a note app, plans in another tool, agents across a pile of terminal windows, results in a git client. Editor-centric development has the IDE; CLI-agent-centric development doesn't — so the one ferrying context between tools ends up being you.
|
|
48
48
|
>
|
|
49
|
-
>
|
|
49
|
+
> CLITrigger is that missing IDE. It connects the whole workflow into a single five-stage pipeline — build project knowledge in **Docs**, shape it into a plan with the **planner & calendar**, refine it live in **terminal sessions**, hand it to multiple AI CLIs (**Claude Code · Antigravity · Codex**) for **parallel autonomous execution** in isolated git worktrees, and land the results through the **review queue and built-in Git client**.
|
|
50
50
|
>
|
|
51
|
-
> **
|
|
51
|
+
> **Each stage inherits the context of the one before it — the intent you captured in docs flows all the way to the merge.**
|
|
52
52
|
|
|
53
53
|
```mermaid
|
|
54
54
|
flowchart LR
|
|
@@ -68,20 +68,16 @@ flowchart LR
|
|
|
68
68
|
|
|
69
69
|
## Why CLITrigger?
|
|
70
70
|
|
|
71
|
-
**
|
|
71
|
+
**The tools don't talk to each other.** As AI writes more of the code, the developer's real job becomes capturing intent and reviewing output — yet that intent lives in a note app, the plan in another tool, the agents in a stack of terminal windows, and the results in a git client. Editor-centric development solved this decades ago with the IDE. CLI-agent-centric development never got one.
|
|
72
72
|
|
|
73
|
-
|
|
73
|
+
CLITrigger is built as that IDE. Its backbone is a single pipeline — **Docs → Plan → Terminal → Autonomous Tasks → Version Control** — where each stage is not a separate tool but a consumer of the previous stage's context: docs become plans, plans become the agent's prompt, execution results arrive in a review queue. Intent is never lost between stages.
|
|
74
74
|
|
|
75
|
-
|
|
75
|
+
Inside the pipeline sits the execution machinery:
|
|
76
76
|
|
|
77
|
-
|
|
78
|
-
|
|
79
|
-
|
|
80
|
-
|
|
81
|
-
- **Right now** — Multiple tasks run in isolated git worktrees, with Claude / Antigravity / Codex executing in parallel
|
|
82
|
-
- **Without hitting limits** — Schedule tasks for off-peak hours to make the most of your token quota
|
|
83
|
-
- **Without losing the thread** — The intent captured in your docs flows through plan → terminal → autonomous tasks → merge as one pipeline. No sticky notes, no twelve browser tabs
|
|
84
|
-
- **Better output** — Multiple AI agents debate and review before implementation, producing higher-quality results than a single AI working alone
|
|
77
|
+
- **Parallel execution** — every task runs in its own isolated git worktree, with Claude / Antigravity / Codex working simultaneously
|
|
78
|
+
- **Scheduling around rate limits** — cron-based runs and auto-retry at quota reset make full use of your tokens, even while you're away
|
|
79
|
+
- **Multi-agent quality** — architect / developer / reviewer agents debate before implementation, beating a single agent working alone
|
|
80
|
+
- **One place to land it** — triage every diff in the review queue, then commit, push, and merge in the built-in Git client
|
|
85
81
|
|
|
86
82
|
---
|
|
87
83
|
|
package/README_KR.md
CHANGED
|
@@ -6,9 +6,9 @@
|
|
|
6
6
|
<img alt="CLITrigger" src="https://raw.githubusercontent.com/HyperAITeam/CLITrigger/main/src/client/public/logo.svg" width="360">
|
|
7
7
|
</picture>
|
|
8
8
|
|
|
9
|
-
**AI
|
|
9
|
+
**AI CLI 에이전트를 위한 IDE**
|
|
10
10
|
|
|
11
|
-
|
|
11
|
+
*문서 · 계획 · 터미널 · 자동작업 · git — 흩어진 도구 다섯 개 대신, 하나의 워크스페이스.*
|
|
12
12
|
|
|
13
13
|
<p align="center">
|
|
14
14
|
<a href="https://github.com/HyperAITeam/CLITrigger/blob/main/README.md">English</a> ·
|
|
@@ -44,11 +44,11 @@ npm i -g clitrigger && clitrigger
|
|
|
44
44
|
|
|
45
45
|
> ### 문서 → 계획 → 터미널 → 자동작업 → 형상관리. 하나의 파이프라인.
|
|
46
46
|
>
|
|
47
|
-
>
|
|
47
|
+
> AI CLI 에이전트(Claude Code, Codex 같은 도구)로 개발하면 도구가 전부 따로 논다 — 요구사항은 노트 앱에, 작업 계획은 또 다른 툴에, 에이전트는 터미널 창 여러 개에, 결과물은 git 클라이언트에. 에디터 중심 개발에는 IDE라는 통합 환경이 있지만 CLI 에이전트 중심 개발에는 그런 게 없었고, 매번 도구 사이에서 맥락을 복사해 나르는 건 결국 사람이다.
|
|
48
48
|
>
|
|
49
|
-
>
|
|
49
|
+
> CLITrigger는 그 빈자리를 채우는 IDE다. 개발 전체를 다섯 단계의 하나의 파이프라인으로 잇는다 — 프로젝트 지식을 **문서**에 쌓고, **플래너·캘린더**로 계획을 세우고, **터미널 세션**에서 AI와 함께 다듬고, 확정된 작업은 격리된 git worktree에서 여러 AI CLI(**Claude Code · Antigravity · Codex**)가 **병렬로 자동 실행**하고, 결과는 **리뷰 큐와 내장 Git 클라이언트**에서 머지까지 끝낸다.
|
|
50
50
|
>
|
|
51
|
-
>
|
|
51
|
+
> **각 단계는 앞 단계의 맥락을 그대로 이어받는다 — 문서에 캡처한 의도가 유실 없이 머지까지 흐른다.**
|
|
52
52
|
|
|
53
53
|
```mermaid
|
|
54
54
|
flowchart LR
|
|
@@ -68,20 +68,16 @@ flowchart LR
|
|
|
68
68
|
|
|
69
69
|
## 왜 CLITrigger인가?
|
|
70
70
|
|
|
71
|
-
|
|
71
|
+
**도구가 전부 따로 논다.** AI가 코드를 더 많이 짤수록 개발자의 진짜 역할은 **의도를 캡처하고 결과를 리뷰하는 것**이 된다 — 그런데 그 의도는 노트 앱에, 계획은 또 다른 툴에, 에이전트는 터미널 창 여러 개에, 결과물은 git 클라이언트에 흩어져 있다. 에디터 중심 개발은 IDE라는 통합 환경으로 이 문제를 진작에 풀었지만, CLI 에이전트 중심 개발에는 그런 게 없었다.
|
|
72
72
|
|
|
73
|
-
|
|
73
|
+
CLITrigger는 그 IDE로 설계됐다. 뼈대는 하나의 파이프라인 — **문서 → 계획 → 터미널 → 자동작업 → 형상관리** — 이고, 각 단계는 독립된 도구가 아니라 앞 단계의 맥락을 이어받는다. 문서가 계획이 되고, 계획이 에이전트의 프롬프트가 되고, 실행 결과가 리뷰 큐에 도착한다. 의도(intent)가 단계 사이에서 유실되지 않는다.
|
|
74
74
|
|
|
75
|
-
|
|
75
|
+
파이프라인 안에는 실행을 받치는 장치들이 있다:
|
|
76
76
|
|
|
77
|
-
|
|
78
|
-
|
|
79
|
-
|
|
80
|
-
|
|
81
|
-
- **지금 당장** — 여러 작업을 격리된 worktree에서 Claude / Antigravity / Codex가 병렬로 처리
|
|
82
|
-
- **한도 걱정 없이** — 새벽, 특정 시각에 예약 실행으로 토큰을 최대한 활용
|
|
83
|
-
- **맥락을 잃지 않고** — 문서에 캡처한 의도가 계획 → 터미널 → 자동작업 → 머지까지 한 파이프라인으로 흐른다. 포스트잇도, 탭 열두 개도 필요 없다
|
|
84
|
-
- **더 나은 결과** — 여러 AI 에이전트가 서로 토론한 뒤 구현, 혼자 짠 코드보다 품질이 높아진다
|
|
77
|
+
- **병렬 실행** — TODO마다 격리된 git worktree에서 Claude / Antigravity / Codex가 동시에 작업
|
|
78
|
+
- **한도를 피하는 예약 실행** — cron 스케줄과 rate limit 리셋 시각 자동 재시도로, 자리를 비운 동안에도 토큰을 끝까지 활용
|
|
79
|
+
- **멀티 에이전트 품질** — 아키텍트·개발자·리뷰어 에이전트가 구현 전에 먼저 토론, 혼자 짠 코드보다 품질이 높아진다
|
|
80
|
+
- **한 곳에서 마무리** — 리뷰 큐에서 diff를 트리아지하고, 내장 Git 클라이언트에서 커밋·푸시·머지까지
|
|
85
81
|
|
|
86
82
|
---
|
|
87
83
|
|