drjmf RagFlow Agent Report Skill
This skill extracts and analyzes RagFlow logs and configurations to provide a detailed report on agent performance, tool usage, metadata filtering, and execution timing.
Workflow
- Identify Agent ID: Ensure you have the target RagFlow agent ID (e.g.,
bd62e008795d11f1a2a4f704f892ca8d). - Identify Timezone: Check if the user wants UTC (log default) or a local timezone like Lisbon (UTC+1).
- Run Diagnostic Script: Use the bundled Python script to map tools to datasets and fetch recent logs.
- Command:
python3 /root/.gemini/skills/ragflow-agent-report/scripts/report.py <agent_id>
- Command:
- Extract Metadata Config: Check the Agent DSL in the database to find:
- Metadata field names (e.g.,
literacy_component,group_level_id). - Filter operators (e.g.,
contains,equal).
- Metadata field names (e.g.,
- Analyze Logs: Extract the execution timeline and calculate durations:
- Metadata filter generation duration.
- Tool call durations.
- Answer compilation time.
- Present Report: Create a structured table including Metadata Tags and Durations.
Resources
- Report Script:
/root/.gemini/skills/ragflow-agent-report/scripts/report.py- Fetches tool mappings and log tails from MySQL and Docker.
Example Report Structure
Internal Execution Steps
| Step | Time (UTC) | Action / Tool Used | Applied Metadata Tags | Result | Duration |
|---|---|---|---|---|---|
| 1 | HH:MM:SS | Tool Call: search_my_dateset_N | tag_name: value |
Found: Document X | Xs |
Tool to Dataset Mapping
| Tool ID | Alias | Dataset Name | Metadata Fields Used |
|---|---|---|---|
search_my_dateset_7 |
Retrieval_Manuals |
Credo Teaching Manuals | literacy_component, language |