Skip to content

Commit 87e9abd

Browse files
author
Focaxis Dev
committed
Improve README positioning
1 parent dca820a commit 87e9abd

2 files changed

Lines changed: 20 additions & 6 deletions

File tree

README.md

Lines changed: 19 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -1,22 +1,36 @@
11
# Deja Vu
22

3-
**A cue-first memory protocol for AI agents**
3+
**Stop AI agents from forgetting project context without adding a database, vector store, or memory service.**
44

5-
Deja Vu is a lightweight memory protocol built around one loop:
5+
Deja Vu is an AI agent memory protocol for teams that want durable project context in ordinary repo files.
6+
7+
Most agent memory tools start by storing more text. Deja Vu starts by asking a cheaper question: does this task feel familiar enough to justify loading memory at all?
8+
9+
```text
10+
before: agent reads too much, forgets why decisions were made, or carries noisy transcript history
11+
after: agent scans tiny cues, loads only relevant memory, and writes back durable project knowledge
12+
```
13+
14+
The core loop is intentionally small:
615

716
```text
817
task cue -> familiarity score -> minimal recall -> durable writeback
918
```
1019

11-
The protocol is packaged as three project-local assets:
20+
It is packaged as three project-local assets:
1221

1322
- rules
1423
- workflow
1524
- tiny memory files
1625

17-
The goal is not to give every agent a heavy runtime. The goal is to give any agent a repeatable discipline for spending almost no tokens until the task proves that deeper memory is useful.
26+
No server. No hidden state. No required npm package. The base protocol works with Markdown and JSONL files that live beside the code.
27+
28+
Use Deja Vu when you want:
1829

19-
The current patch line also emphasizes memory quality control: compact cue linting, gist-first summaries, and boundary-aware chunks keep recall routes small instead of merely adding more stored text.
30+
- project memory that survives a new chat or agent session
31+
- low-token recall before planning, coding, or answering
32+
- durable decisions, preferences, open loops, and architecture intent in plain files
33+
- an optional TypeScript semantic engine only when the project outgrows the file-first protocol
2034

2135
## What Deja Vu Is
2236

encoding-status.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -8,7 +8,7 @@
88
| `package-lock.json` | 編碼正常(已檢查) | Updated in UTF-8; package version metadata now matches the 0.3.1 release. |
99
| `tsconfig.json` | 編碼正常(新建,UTF-8) | New text file. |
1010
| `LICENSE` | 編碼正常(新建,UTF-8) | New text file. |
11-
| `README.md` | 編碼正常(已檢查) | Updated in UTF-8; overview now includes cue quality control, gist summaries, and boundary-aware chunks. |
11+
| `README.md` | 編碼正常(已檢查) | Updated in UTF-8; GitHub-facing opening now emphasizes agent context loss, no database/vector dependency, and before/after value. |
1212
| `CHANGELOG.md` | 編碼正常(已檢查) | Updated in UTF-8; now includes the 0.3.1 recall-quality patch release notes. |
1313
| `llms.txt` | 編碼正常(已檢查) | Updated in UTF-8; AI-readable index now includes cue quality linting, gist summaries, and boundary-aware chunking. |
1414
| `docs/architecture.md` | 編碼正常(已檢查) | Rewritten in UTF-8; architecture doc now describes the engine as the optional layer inside a protocol-first product. |

0 commit comments

Comments
 (0)