refactor: rename mem_scanner→procmem_scanner, add L4 layer, remove make_prompts.py
This commit is contained in:
@@ -20,6 +20,7 @@ L1: global_mem_insight.txt (极简索引层 - 严格控制 ≤30 行)
|
||||
L2: global_mem.txt (事实库层 - 现短但会膨胀)
|
||||
↓ 详细引用 (Reference)
|
||||
L3: ../memory/ (记录库层 - 包含 .md/.py 等各类文件)
|
||||
L4: ../memory/L4_raw_sessions/ (历史会话层 - scheduler反射自动收集,可定位过往上下文)
|
||||
```
|
||||
---
|
||||
## 各层职责与原则
|
||||
|
||||
@@ -7,7 +7,7 @@
|
||||
```python
|
||||
import sys
|
||||
sys.path.append('../memory') # 直接挂载工具目录
|
||||
from mem_scanner import scan_memory
|
||||
from procmem_scanner import scan_memory
|
||||
|
||||
# 示例:搜索特定 Hex 特征码,开启 llm_mode 以获取上下文
|
||||
results = scan_memory(pid, "48 8b ?? ?? 00", mode="hex", llm_mode=True)
|
||||
@@ -16,10 +16,10 @@ results = scan_memory(pid, "48 8b ?? ?? 00", mode="hex", llm_mode=True)
|
||||
**CLI:**
|
||||
```powershell
|
||||
# 基础搜索
|
||||
python ../memory/mem_scanner.py <PID> "pattern" --mode string
|
||||
python ../memory/procmem_scanner.py <PID> "pattern" --mode string
|
||||
|
||||
# LLM 增强模式(输出包含上下文的 JSON,推荐)
|
||||
python ../memory/mem_scanner.py <PID> "pattern" --llm
|
||||
python ../memory/procmem_scanner.py <PID> "pattern" --llm
|
||||
```
|
||||
|
||||
## 2. 典型场景:结构体或关键数据定位
|
||||
@@ -9,7 +9,7 @@
|
||||
- 流程:启动 → 轮询 output.txt(`[ROUND END]`=该轮完成)→ 写 reply.txt 继续 → 不写则10min退出
|
||||
- 干预文件:`_stop`(当轮结束退出) | `_keyinfo`(写入即注入下轮prompt的key_info) | `_intervene`(写入内容作为下轮追加指令)
|
||||
- output1/2/3.txt:reply后各轮输出(递增编号),同样append+`[ROUND END]`
|
||||
- input.txt:目标+约束,可指定SOP名。禁写具体步骤(除非已读SOP确认)。大量数据给路径禁塞入
|
||||
- input.txt:目标+约束,可指定SOP名。**禁写具体步骤**。大量数据给路径禁塞入
|
||||
- ⚠ `--input`走命令行,长文本/含引号会超时。超过一句话时:先写input.txt,启动时不带`--input`
|
||||
- --bg启动瞬间返回,可同一code_run内sleep后poll;非--bg方式禁止合并启动+轮询
|
||||
|
||||
|
||||
Reference in New Issue
Block a user