Skip to content
Open
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion docs/INTERNALS.md
Original file line number Diff line number Diff line change
Expand Up @@ -127,7 +127,7 @@ Ombre-Brain/
- **reclassify_api.py** — 一次性脚本:把历史落在「未分类/」的桶重新 `analyze()` 打标并搬到正确 domain 目录,只改 frontmatter 与文件位置。
- **errors.py** — OB 统一错误码(如 OB-W005 embedding 模型漂移、OB-Startup 系列),供各模块抛结构化异常。

> **第一人称豁免**:`import_memory.py` 在喂给 LLM 的 prompt 里把对话格式化成 `[用户] ... [AI] ...` 文本块([src/import_memory.py](src/import_memory.py) `_chunk_turns` 第 291 行),这是给 LLM 看的「对话块」标签,不是写入桶 frontmatter 或返回给模型的 docstring,因此不违反 §2.9 第一人称原则。修改这段时勿误删。
> **第一人称豁免**:`import_memory.py` 在喂给 LLM 的 prompt 里把对话格式化成 `[用户] ... [AI] ...` 文本块([src/import_memory.py](../src/import_memory.py) 的 `chunk_turns`),这是给 LLM 看的「对话块」标签,不是写入桶 frontmatter 或返回给模型的 docstring,因此不违反 §2.9 第一人称原则。修改这段时勿误删。
> **导入阈值**:`_PATTERN_MIN_DYNAMIC_BUCKETS = 5` / `_PATTERN_PIN_SUGGEST_THRESHOLD = 5`,详见 rule.md §6 备注。
- **utils.py** — 配置加载(env > yaml > defaults 三级优先级)、日志、12 位 hex 短 ID 生成、`safe_path()` 路径遍历防护、`count_tokens_approx()` 中英混排 token 估算。

Expand Down