Back to Details

vector-db-search

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Vector DB Search

向量数据库搜索

Semantic (meaning-based) search against the ChromaDB vector store. Use for Phase 2 of the 3-phase search protocol -- after the RLM Summary Ledger (Phase 1) returns insufficient results.
基于语义（即基于含义）的ChromaDB向量存储搜索。 用于三步搜索协议的第二阶段——当RLM摘要分类账（第一阶段）返回的结果不足时使用。

Scripts

脚本

ScriptRole
scripts/query.py
Semantic search -- CLI entry point
scripts/operations.py
Core Parent-Child retrieval library
scripts/vector_config.py
Profile config helper (
vector_profiles.json
)
scripts/vector_consistency_check.py
Integrity validation
Write operations (ingest, cleanup) are handled by dedicated agents: 
vdb-ingest
, 
vdb-cleanup
.
脚本作用
scripts/query.py
语义搜索——CLI入口点
scripts/operations.py
核心父子检索库
scripts/vector_config.py
配置文件辅助工具（
vector_profiles.json
）
scripts/vector_consistency_check.py
完整性验证
写入操作（如数据导入、清理）由专用Agent处理：
vdb-ingest
、
vdb-cleanup
。

When to Use

使用场景

  • Phase 1 (RLM Summary Ledger) returned no match or insufficient detail
  • User asks "how does X work?" / "find code that does Y"
  • You need specific snippets, not just file-level summaries
  • 第一阶段（RLM摘要分类账）未找到匹配结果或细节不足
  • 用户询问"X是如何工作的？" / "查找实现Y功能的代码"
  • 需要特定代码片段，而非仅文件级摘要

Execution Protocol

执行流程

1. Verify ChromaDB is running

1. 验证ChromaDB是否运行

bash
curl -sf http://127.0.0.1:8110/api/v1/heartbeat
If connection refused: run 
vector-db-launch
skill (
plugins/vector-db/skills/vector-db-launch/SKILL.md
). For first-time setup: run 
vector-db-init
skill (
plugins/vector-db/skills/vector-db-init/scripts/init.py
).
bash
curl -sf http://127.0.0.1:8110/api/v1/heartbeat
如果连接被拒绝：运行
vector-db-launch
Skill（
plugins/vector-db/skills/vector-db-launch/SKILL.md
）。 首次设置：运行
vector-db-init
Skill（
plugins/vector-db/skills/vector-db-init/scripts/init.py
）。

2. Select Profile and Search

2. 选择配置文件并搜索

Profiles are project-defined in 
vector_profiles.json
(see 
vector-db-init
skill). Any number can exist. Discover what's available:
bash
cat .agent/learning/vector_profiles.json
Common default is 
knowledge
-- your project may define more (e.g. separate profiles for code vs docs). When topic is ambiguous, search all profiles.
bash
python3 plugins/vector-db/skills/vector-db-agent/scripts/query.py \
  "your natural language question" --profile knowledge --limit 5
Results include ranked parent chunks with RLM Super-RAG context pre-injected.
配置文件在
vector_profiles.json
中由项目定义（请查看
vector-db-init
Skill），可存在任意数量。查看可用配置文件：
bash
cat .agent/learning/vector_profiles.json
常见默认配置文件为
knowledge
——你的项目可能定义了更多（例如，代码和文档使用单独的配置文件）。当主题不明确时，搜索所有配置文件。
bash
python3 plugins/vector-db/skills/vector-db-agent/scripts/query.py \
  "your natural language question" --profile knowledge --limit 5
结果包含已注入RLM Super-RAG上下文的排序父块。

Architectural Constraints (Electric Fence)

架构约束（安全边界）

NEVER -- direct database reads

禁止——直接读取数据库

Do not 
cat
, 
strings
, or 
sqlite3
the 
.vector_data/
directory. Binary blobs will corrupt your context window and the retrieval pipeline.
Do not 
cat
, 
strings
, or 
sqlite3
the 
.vector_data/
directory. Binary blobs will corrupt your context window and the retrieval pipeline.

ALWAYS -- use the API

必须——使用API

All access goes through 
query.py
. No exceptions.
All access goes through 
query.py
. No exceptions.

Source Transparency Declaration (L5 Pattern)

源透明声明（L5模式）

When search returns empty results, explicitly state:
> Not Found in Vector Store
> Searched profile: [profile_name] for "[query]"
> Profile covers: [scope]
> Not searched: [out-of-scope areas]
当搜索返回空结果时，需明确声明：
> Not Found in Vector Store
> Searched profile: [profile_name] for "[query]"
> Profile covers: [scope]
> Not searched: [out-of-scope areas]