Ragflow Performance Report

RagFlow Performance Report

This skill automates the identification of performance bottlenecks in RagFlow agents. It specifically targets issues that cause high latency (90s+) and massive token overhead.

Workflow

1. Identify Agent ID: Obtain the agent ID from the user (e.g., 4e7fe12c665211f1815b89b948263ffa).
2. Run Analysis Script: Execute the bundled Python script to pull the Agent DSL from the database and analyze its configuration.
   • Command: python3 /root/.gemini/skills/ragflow-performance-report/scripts/analyze.py <agent_id>
3. Analyze Findings:
   • TOC Enhancement: Check if toc_enhance is enabled on large manuals. This is the #1 cause of "1M+ token" context bloat.
   • Top_K Settings: Check for top_k > 512 which can slow down vector retrieval.
   • Rerankers: Check if a reranker is missing, which often forces users to over-rely on TOC enhancement for quality.
4. Present Recommendations: Provide a structured list of critical fixes and optimizations.

Resources

• Analysis Script: /root/.gemini/skills/ragflow-performance-report/scripts/analyze.py
  • Queries rag_flow.user_canvas or rag_flow.agent tables.
  • Extracts component parameters for all retrieval tools.