技能 cocoindex
🔄

cocoindex

安全

Build AI data pipelines with CocoIndex

也可從以下取得: Joseph OBrien,davila7

Creating data transformation pipelines for AI applications requires understanding complex ETL patterns, embedding models, and vector databases. CocoIndex provides a unified framework for building real-time indexing flows that extract from multiple sources, transform with chunking and embeddings, and export to vector databases and knowledge graphs.

支援: Claude Codex Code(CC)
🥉 72 青銅
1

下載技能 ZIP

2

在 Claude 中上傳

前往 設定 → 功能 → 技能 → 上傳技能

3

開啟並開始使用

測試它

正在使用「cocoindex」。 Build a CocoIndex flow that embeds my documents

預期結果:

  • Set up project with cocoindex package
  • Create flow definition with LocalFile source
  • Apply SplitRecursively for chunking
  • Use SentenceTransformerEmbed or EmbedText for vectors
  • Export to vector database target
  • Run setup then update to build index

安全審計

安全
v3 • 1/10/2026

Pure documentation skill containing only markdown reference files for the CocoIndex library. No executable code, scripts, or runtime components. This skill only displays documentation and does not perform any file access, network operations, or code execution.

5
已掃描檔案
1,640
分析行數
0
發現項
3
審計總數
未發現安全問題
審計者: claude 查看審計歷史 →

品質評分

45
架構
100
可維護性
83
內容
30
社群
100
安全
87
規範符合性

你能建構什麼

Build vector search indexes

Create pipelines that embed documents and store in vector databases for semantic search.

Process data for AI applications

Transform raw data through chunking, embedding, and extraction for AI model consumption.

Construct knowledge graphs

Extract structured entities using LLMs and build graph databases for relationship-based queries.

試試這些提示

Create vector index 
Help me create a CocoIndex flow that reads markdown files from a local directory, splits them into chunks of 2000 characters with 500 overlap, generates embeddings using OpenAI text-embedding-3-small, and exports to Postgres with pgvector for semantic search.
Build knowledge graph 
Show me how to use CocoIndex to read JSON product files, extract structured information using GPT-4, and export the results as nodes and relationships in a Neo4j knowledge graph.
Implement live updates 
I want to create a CocoIndex flow with live updates. Help me configure a local file source with a refresh interval and set up automatic processing when files change.
Write custom function 
I need to create a custom CocoIndex function that calls an external API to enrich my data. Show me how to use the spec+executor pattern with caching and API authentication.

最佳實務

  • Use evaluate command to test flows before running update
  • Always assign transformed data to row fields, not local variables
  • Increment behavior_version when modifying cached functions
  • Add refresh_interval to sources for live update mode

避免

  • Using local variables instead of row fields for transformation results
  • Creating unnecessary dataclasses to mirror flow field schemas
  • Omitting type annotations on custom function return values
  • Running update without first running setup on new flows

常見問題

Which AI tools is CocoIndex compatible with?
CocoIndex works with OpenAI, Anthropic, Gemini, Voyage, and Ollama for embeddings and LLM extraction. Claude, Codex, and Claude Code can all use CocoIndex flows.
What are the size limits for data processing?
CocoIndex supports configurable concurrency limits. Set max_inflight_rows and max_inflight_bytes to control memory usage during processing.
How do I integrate with my existing codebase?
Install cocoindex package, define flows as Python functions with @cocoindex.flow_def decorator, then use CLI or Python API to operate flows.
Is my data safe when using CocoIndex?
CocoIndex runs locally with your data. API keys are read from environment variables. Source data stays on your machine except when explicitly exported to configured targets.
Why does my flow fail with database connection error?
Ensure COCOINDEX_DATABASE_URL is set in your .env file. The default is postgres://cocoindex:cocoindex@localhost/cocoindex for local development.
How does CocoIndex compare to LangChain or LlamaIndex?
CocoIndex focuses on real-time ETL pipelines with incremental processing. It complements orchestration libraries by handling the indexing and data transformation layer.

開發者詳情

授權

MIT

儲存庫

https://github.com/89jobrien/steve/tree/main/steve/skills/cocoindex

引用

main

檔案結構

📁 references/

📄 api-operations.md

📄 cli-operations.md

📄 custom-functions.md

📄 flow-patterns.md

📄 SKILL.md