Bỏ qua tới nội dung
TakoAPI
Quay lại kỹ năng

snipara-mcp

bởi alopez3006

Coding Agents & IDEs
1.4kLượt tải do clawskills.sh công bố
1Sao
0Lượt xem
0 lượt thíchĐăng nhập để thích kỹ năng này
—Chưa có đánh giá
Đánh giá của bạn

Tóm tắt

You are an AI assistant with access to Snipara MCP tools.

Mô tả

Kỹ năng này làm gì

Snipara MCP adds semantic documentation search to AI assistants. It queries pre-indexed docs instead of reading files one by one, cutting context usage and query time. Supports cross-repo search and optional persistent memory across sessions.

Pre-indexed semantic search returns relevant context in 2 seconds vs 20+ seconds of sequential file reads that can exhaust token limits before finding an answer.

Khi nào nên dùng

  • Finding how an auth system works without reading dozens of source files
  • Searching webhook implementations across multiple microservices simultaneously
  • Storing coding preferences so AI recalls them in future sessions
  • Auto-injecting team coding standards into every AI query
  • Reducing token usage when answering codebase architecture questions

Quy trình ví dụ

ĐẦU VÀO

User asks: 'How does our authentication system work?'

AGENT
  1. 1

    Call rlm_ask('authentication system') to query indexed docs

  2. 2

    Receive top relevant documentation sections in about 2 seconds

  3. 3

    If insufficient, call rlm_context_query with search_mode set to semantic

  4. 4

    Synthesize answer from returned context, using 3K tokens instead of 50K

ĐẦU RA

A clear explanation of the authentication system sourced from pre-indexed docs

Yêu cầu

  • Snipara account (snipara.com, free tier: 100 queries/month)
  • SNIPARA_API_KEY environment variable
  • Documentation files uploaded and indexed on Snipara dashboard
  • Pro plan ($19/mo) for semantic search or Team plan ($49/mo) for multi-repo; separate Agents plan required for memory features

Cài đặt

clawhub install alopez3006/snipara-mcp

Chạy lệnh CLI, hoặc dán prompt vào Claude Code, Cursor, Codex hay coding agent bạn đang dùng.

Liên kết