Skip to content

Commit ed3dbd0

Browse files
authored
feat: review contents (#17)
* feat: review contents * chore: save some md files but they do not publish
1 parent a679622 commit ed3dbd0

25 files changed

Lines changed: 2662 additions & 361 deletions

docs/LONG_CONTEXT.md

Lines changed: 38 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,38 @@
1+
# Generate LONG_CONTEXT.md
2+
3+
지금까지의 대화 내용을 기반으로 LONG_CONTEXT.md를 생성해줘.
4+
5+
LONG_CONTEXT.md의 목적은 새로운 Session 또는 새로운 Model이 현재 상황을 빠르게 복구할 수 있도록 하는 것이다.
6+
7+
이 문서는 프로젝트 문서가 아니다.
8+
9+
다음 목적을 가진 Session Bootstrap 문서다.
10+
11+
- 새로운 Session 시작
12+
- 새로운 Model 사용
13+
- 장기간 중단 후 프로젝트 재개
14+
- 채팅 Context 소실 복구
15+
16+
작성 원칙:
17+
18+
1. 대화 내용을 단순 요약하지 말 것
19+
2. 프로젝트의 현재 상태를 설명할 것
20+
3. 중요한 의사결정과 그 이유를 기록할 것
21+
4. 논의 과정에서 얻은 핵심 통찰을 기록할 것
22+
5. 앞으로 이어서 논의해야 할 사항을 기록할 것
23+
6. 현재 사용자가 중요하게 생각하는 철학과 원칙을 기록할 것
24+
7. AGENTS.md, ARCHITECTURE.md, TASKS.md에 이미 존재하는 내용을 장황하게 복사하지 말 것
25+
8. 새로운 모델이 "왜 이런 구조가 되었는가"를 이해할 수 있도록 작성할 것
26+
9. 세부 구현보다 의사결정 배경을 우선 기록할 것
27+
28+
포함 항목:
29+
30+
- Current Project
31+
- Current Status
32+
- Key Decisions
33+
- Important Insights
34+
- Open Questions
35+
- Current Direction
36+
- Next Discussion Topics
37+
38+
출력은 LONG_CONTEXT.md 전체 문서만 작성할 것.

docs/MANIFESTO.mdx

Lines changed: 100 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,100 @@
1+
---
2+
title: 'REPL Works Manifesto'
3+
date: 2026-06-06
4+
---
5+
6+
## Introduction
7+
8+
We are discovering better ways to build products with AI.
9+
10+
AI models generate better code every year.
11+
12+
Yet long-term projects continue to fail for familiar reasons.
13+
14+
Project intent disappears.
15+
16+
Architecture drifts.
17+
18+
The same mistakes repeat.
19+
20+
Context is lost between sessions.
21+
22+
Changing models often means starting over.
23+
24+
We do not believe this happens because AI lacks intelligence.
25+
26+
We believe it happens because projects lack memory.
27+
28+
---
29+
30+
## Core Belief
31+
32+
AI memory is temporary.
33+
34+
Project memory must be externalized.
35+
36+
Models forget.
37+
38+
Projects must not.
39+
40+
---
41+
42+
## Values
43+
44+
We value Project Memory over Model Memory.
45+
46+
We value Architecture over Prompt Engineering.
47+
48+
We value Persistent Knowledge over Long Conversations.
49+
50+
We value Workflow Consistency over Agent Intelligence.
51+
52+
We value Model Independence over Single Model Dependency.
53+
54+
We value Sustainable Development over Fast Implementation.
55+
56+
---
57+
58+
## Principles
59+
60+
### Project knowledge must exist outside the model
61+
62+
Projects must survive model upgrades, vendor changes, and new sessions.
63+
64+
### Project memory must be persistent
65+
66+
Knowledge that matters should not disappear when a conversation ends.
67+
68+
### Architecture defines intent
69+
70+
Implementation should follow architecture, not replace it.
71+
72+
### Context is a limited resource
73+
74+
Important context must be preserved, not repeatedly reconstructed.
75+
76+
### Models are replaceable
77+
78+
Project continuity should not depend on a specific model.
79+
80+
### Long-term development requires long-term memory
81+
82+
Projects live longer than sessions.
83+
84+
Memory must live longer than context windows.
85+
86+
---
87+
88+
## Final Statement
89+
90+
We are not trying to make AI smarter.
91+
92+
We are trying to make projects more durable.
93+
94+
The future of AI development is not better memory inside models.
95+
96+
It is better memory outside them.
97+
98+
Models forget.
99+
100+
Projects must not.

0 commit comments

Comments
 (0)