emdb-insert
Insert vectors into EmergentDB using the SDK. Use when the user wants to store embeddings, index documents, or batch upload vectors into EmergentDB.
ソース情報
- リポジトリ
- justrach/emergent-sdk
- ソースの最終更新活動
- 2026年2月23日 18:35
- 検出された SKILL.md の言語
- 英語
- スター
- 0
- フォーク
- 0
インストール方法
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
ソースファイルを確認
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
SKILL.md を表示中
SKILL.md
ソースの指示 · 読み取り専用プレビュー- name
- emdb-insert
- description
- Insert vectors into EmergentDB using the SDK. Use when the user wants to store embeddings, index documents, or batch upload vectors into EmergentDB.
- allowed-tools
- Bash, Read, Write, Edit
# Insert Vectors into EmergentDB
Help the user insert vectors into EmergentDB using the official SDKs.
## Prerequisites
- An EmergentDB API key (starts with `emdb_`)
- Vector embeddings (default dimension: 1536 for OpenAI text-embedding-3-small)
## TypeScript SDK
```typescript
import { EmergentDB } from "emergentdb";
const db = new EmergentDB("emdb_your_api_key");
// Single insert
const result = await db.insert(1, embedding, { title: "My doc" }, "production");
// result: { success: true, id: 1, namespace: "production", upserted: false }
// Batch insert (max 1000 per call)
const batch = await db.batchInsert([
{ id: 1, vector: [...], metadata: { title: "Doc 1" } },
{ id: 2, vector: [...], metadata: { title: "Doc 2" } },
], "production");
// Unlimited batch (auto-chunks into 1000s)
const all = await db.batchInsertAll(vectors, "production");
```
## Python SDK
```python
from emergentdb import EmergentDB
db = EmergentDB("emdb_your_api_key")
# Single insert
result = db.insert(1, embedding, metadata={"title": "My doc"}, namespace="production")
# Batch insert
result = db.batch_insert([
{"id": 1, "vector": [...], "metadata": {"title": "Doc 1"}},
{"id": 2, "vector": [...], "metadata": {"title": "Doc 2"}},
], namespace="production")
# Unlimited batch
result = db.batch_insert_all(vectors, namespace="production")
```
## Key Behaviors
- **Upsert**: Re-inserting an existing ID in the same namespace replaces it. Response shows `upserted: true`.
- **Namespaces**: Optional, defaults to `"default"`. Created automatically on first use.
- **Metadata fields**: `title` (str), `content` (str), `source_url` (str), `tags` (str[])
- **Batch limit**: Max 1,000 vectors per `batchInsert` / `batch_insert` call. Use `batchInsertAll` / `batch_insert_all` for larger sets.
- **Dimensions**: Must match the index dimension (1536 by default). Mismatched dimensions return HTTP 400.
## Error Codes
| Code | Meaning |
|------|---------|
| 400 | Invalid request — bad JSON, wrong vector dimension, missing field |
| 401 | Missing or invalid API key |
| 402 | Vector capacity exceeded — upgrade plan |
| 429 | Rate limit exceeded — back off and retry |
| 500 | Server error — retry with backoff |
## Rate Limits
| Plan | Limit |
|------|-------|
| Free | 60 req/min |
| Launch | 300 req/min |
| Scale | 600 req/min |
## Common Pattern: OpenAI Embeddings
```python
import openai
from emergentdb import EmergentDB
client = openai.OpenAI()
db = EmergentDB("emdb_your_key")
# Generate + store
resp = client.embeddings.create(model="text-embedding-3-small", input="My document text")
db.insert(1, resp.data[0].embedding, metadata={"title": "My document"})
```
When helping the user, determine which SDK they prefer and guide them through the insert flow. If they have raw text, suggest using OpenAI or another embedding model first.
GitHubで見る