🐛 [Bug]: RAG query returns no results
Archon Version
v0.1.0
Bug Severity
🟠 High - Blocks important features
Bug Description
Whatever I do no results are returned with claude code
I've a source https://docs.livekit.io/llms-full.txt all indexed
in claude code
archon - perform_rag_query (MCP)(query: "workflows", source: "docs.livekit.io", match_count: 10)
⎿ {
"success": true,
"results": [],
… +3 lines (ctrl+r to expand)
● archon - perform_rag_query (MCP)(query: "workflows", source: "file_workflows_md_1755555943", match_count: 10)
⎿ {
"success": true,
"results": [],
… +3 lines (ctrl+r to expand)
Docker logs:
Archon-Server | 2025-08-18 22:30:55 | httpx | INFO | HTTP Request: GET https://dinusmhibivlerbyvlwu.supabase.co/rest/v1/archon_crawled_pages?select=id%2Curl%2Cchunk_number%2Ccontent%2Cmetadata%2Csource_id&content=ilike.%25workflow%25&limit=40 "HTTP/2 200 OK"
Archon-Server | 2025-08-18 22:30:55 | src.server.services.search.rag_service | INFO | RAG query completed - 0 results found
Archon-MCP | 2025-08-18 22:30:55 | httpx | INFO | HTTP Request: POST http://archon-server:8181/api/rag/query "HTTP/1.1 200 OK"
Archon-MCP | INFO: 192.168.1.17:51694 - "POST /mcp HTTP/1.1" 200 OK
Archon-MCP | 2025-08-18 22:30:58 | mcp.server.lowlevel.server | INFO | Processing request of type CallToolRequest
Archon-Server | 2025-08-18 22:30:58 | search | INFO | Supabase client initialized - project_id=dinusmhibivlerbyvlwu
Archon-Server | 2025-08-18 22:30:58 | src.server.services.search.reranking_strategy | WARNING | sentence-transformers not available - reranking disabled
Archon-Server | 2025-08-18 22:30:58 | src.server.services.search.rag_service | INFO | Reranking strategy loaded successfully
Archon-Server | 2025-08-18 22:30:58 | src.server.services.search.rag_service | INFO | RAG query started: workflows
Archon-Server | 2025-08-18 22:30:58 | src.server.services.llm_provider_service | INFO | Creating LLM client for provider: openai
Archon-Server | 2025-08-18 22:30:58 | src.server.services.llm_provider_service | INFO | OpenAI client created successfully
Archon-Server | 2025-08-18 22:30:59 | httpx | INFO | HTTP Request: POST https://api.openai.com/v1/embeddings "HTTP/1.1 200 OK"
Archon-Server | 2025-08-18 22:31:00 | httpx | INFO | HTTP Request: POST https://dinusmhibivlerbyvlwu.supabase.co/rest/v1/rpc/match_archon_crawled_pages "HTTP/2 200 OK"
Archon-Server | 2025-08-18 22:31:00 | httpx | INFO | HTTP Request: GET https://dinusmhibivlerbyvlwu.supabase.co/rest/v1/archon_crawled_pages?select=id%2Curl%2Cchunk_number%2Ccontent%2Cmetadata%2Csource_id&content=ilike.%25workflows%25&limit=40 "HTTP/2 200 OK"
Archon-Server | 2025-08-18 22:31:00 | httpx | INFO | HTTP Request: GET https://dinusmhibivlerbyvlwu.supabase.co/rest/v1/archon_crawled_pages?select=id%2Curl%2Cchunk_number%2Ccontent%2Cmetadata%2Csource_id&content=ilike.%25workflow%25&limit=40 "HTTP/2 200 OK"
Archon-Server | 2025-08-18 22:31:00 | src.server.services.search.rag_service | INFO | RAG query completed - 0 results found
Archon-MCP | 2025-08-18 22:31:00 | httpx | INFO | HTTP Request: POST http://archon-server:8181/api/rag/query "HTTP/1.1 200 OK"
Steps to Reproduce
- index "https://docs.livekit.io/llms-full.txt"
- ask claude code anything related to this source
Expected Behavior
Returning any result
Actual Behavior
results []
Error Details (if any)
Affected Component
🔍 Knowledge Base / RAG
Browser & OS
zen
Additional Context
No response
Service Status (check all that are working)
- [x] 🖥️ Frontend UI (http://localhost:3737)
- [x] ⚙️ Main Server (http://localhost:8181)
- [x] 🔗 MCP Service (localhost:8051)
- [x] 🤖 Agents Service (http://localhost:8052)
- [x] 💾 Supabase Database (connected)
Hmmm... not able to reproduce this myself. I crawled the same URL and then asked for a Livekit quickstart. It ran a few different queries such as:
{
match_count: 5,
query: getting started LiveKit tutorial setup install,
source: docs.livekit.io
}
And they all returned results.
Could you check your archon_crawled_pages table in Supabase and see if content seems missing there? Also check the code examples.
Im having the same problem I am crawling the data in archon & it says its successful, but is not adding the data to the crawled_pages table in Supabase. If I create new projects the data is saving to supabase, just seems to be happening for crawled content.
It started working for me automagically I dunno how really I installed archon on a vm on proxmox on my network I gave up last time and powered off the vm and today when I tried again, I added the same sources and It worked even without restarting the containers after adding the sources
I am facing the same problem, This is what my mcp client says { "query": "FastMCP getting started overview installation setup", "source": "gofastmcp.com", "match_count": 10 } Result [ { "type": "text", "text": "{\n "success": true,\n "results": [],\n "reranked": false,\n "error": null\n}" } ]
Yep me too. Never finds anything and is using context7 over knowledge base.
same issue. Last item in logs after crawling (while processing batch) is 2025-08-25 22:02:20 | src.server.services.llm_provider_service | INFO | OpenAI client created successfully
I have the same problem
In logfire I get OPEN_API error despite using gemini.
Gemini project shows no traffic on api key
For people with issues, I do recommend setting up logfire and set log level to DEBUG
Also having the same issue - no results returned, though I can see code samples in archon_code_examples table. Interestingly the archon_crawled_pages and archon_document_versions tables are both empty
FWIW - I dumped the knowledge base, and verified that all tables were empty, then recreated them - using the default OpenAI- LLM and embedings models, and things are now returning.
I will take a look at this one, putting in my todo
I'm having the same issue. I just updated to see if it was resolved but it was not. In my case when I look in supabase all the data is there from the crawls. However, the coding agent never returns results. The output is always. { "success": true, "results": [], "reranked": false, "error": null }
I have this issue as well. Supabase has records in both tables for crawled page and code examples, but rag searches always return 0 results. I'm using Ollama locally with the qwen2.5-coder:latest for LLM instance and mxbai-embed-large:latest for the embedding instance. Not sure if that matters. I always get back something similar to this without any results returned.
archon/rag_search_knowledge_base { "query": "timeslot OR time slot OR availability OR schedule guided tour endpoint list available times appointments tours", "source_domain": "developers.funnelleasing.com", "match_count": 10 }
@dancehands / @frankjoshua and everyone else - the problem right now is the coding assistant passes the URL to filter the query instead of the knowledge source ID. I'm fixing this with this PR that I'll merge to main soon!
#705
Fix for this in main as of today - will be moving this into the first release too!
I pulled in the changes and rebuilt the containers. The issues is still there for me. Although now in the web gui I can see all the documents which I could not before. Every rag search ends in 0 results. Do i need to do any thing else? Like rebuild the knowledge bases.