Skip to content

docker

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

docker

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
Dockerfile.ubuntu
Dockerfile.ubuntu
 
 
README.md
README.md
 
 
build_image.py
build_image.py
 
 
docker-compose.yml
docker-compose.yml
 
 
entrypoint.sh
entrypoint.sh
 
 

README.md

Sirchmunk Docker Images

Available Images

Region Image
US West modelscope-registry.us-west-1.cr.aliyuncs.com/modelscope-repo/sirchmunk:ubuntu22.04-py312-0.0.7
China Beijing modelscope-registry.cn-beijing.cr.aliyuncs.com/modelscope-repo/sirchmunk:ubuntu22.04-py312-0.0.7

Image tag format: ubuntu{ubuntu_version}-py{python_version}-{sirchmunk_version}

Quick Start

1. Pull the image

Choose the registry closest to your location:

# US West
docker pull modelscope-registry.us-west-1.cr.aliyuncs.com/modelscope-repo/sirchmunk:ubuntu22.04-py312-0.0.7

# China Beijing
docker pull modelscope-registry.cn-beijing.cr.aliyuncs.com/modelscope-repo/sirchmunk:ubuntu22.04-py312-0.0.7

2. Start the service

docker run -d \
  --name sirchmunk \
  -p 8584:8584 \
  -e LLM_API_KEY="your-api-key-here" \
  -e LLM_BASE_URL="https://api.openai.com/v1" \
  -e LLM_MODEL_NAME="gpt-5.2" \
  -e LLM_TIMEOUT=60.0 \
  -e UI_THEME=light \
  -e UI_LANGUAGE=en \
  -e SIRCHMUNK_VERBOSE=false \
  -e SIRCHMUNK_SEARCH_PATHS=/mnt/docs \
  -v /path/to/your_work_path:/data/sirchmunk \
  -v /path/to/your/docs:/mnt/docs:ro \
  modelscope-registry.us-west-1.cr.aliyuncs.com/modelscope-repo/sirchmunk:ubuntu22.04-py312-0.0.7

Parameters:

Parameter Required Default Description
-e LLM_API_KEY Yes API key from your LLM provider
-e LLM_BASE_URL No https://api.openai.com/v1 OpenAI-compatible API endpoint (e.g., https://api.minimax.io/v1 for MiniMax)
-e LLM_MODEL_NAME No gpt-5.2 LLM model name (e.g., MiniMax-M2.7, MiniMax-M2.7-highspeed)
-e LLM_TIMEOUT No 60.0 LLM request timeout in seconds
-e UI_THEME No light WebUI theme (light / dark)
-e UI_LANGUAGE No en WebUI language (en / zh)
-e SIRCHMUNK_VERBOSE No false Enable verbose logging (true / false)
-e SIRCHMUNK_SEARCH_PATHS No (empty) Comma-separated default search paths (e.g. /mnt/docs)
-e CHAT_HISTORY_MAX_TURNS No 5 Max chat history turns for multi-turn context
-e CHAT_HISTORY_MAX_TOKENS No 32000 Max tokens for chat history context
-p 8584:8584 Yes Expose WebUI and API port
-v /data/sirchmunk:/data/sirchmunk Recommended Persist data (models, history, knowledge) across restarts

Mount local files for search:

Use -v to mount host directories into the container, then search them via the API or WebUI.

3. Use the service

WebUI — Open http://localhost:8584 in your browser.

API — Search via curl:

curl -X POST http://localhost:8584/api/v1/search \
  -H "Content-Type: application/json" \
  -d '{
    "query": "your search question here",
    "paths": ["/mnt/docs"],
    "mode": "FAST"
  }'

API — Search via Python:

import requests

response = requests.post(
    "http://localhost:8584/api/v1/search",
    json={
        "query": "your search question here",
        "paths": ["/mnt/docs"],
        "mode": "FAST",           # "FAST" (default), "DEEP", or "FILENAME_ONLY"
        "enable_dir_scan": True,  # enable directory scanning (DEEP/FAST)
        "max_depth": 5,           # optional: max directory depth
        "top_k_files": 3,         # optional: number of top files to return
        "max_token_budget": 8192, # optional: LLM token budget (DEEP)
    },
)

result = response.json()
if result["success"]:
    print(result["data"])
else:
    print(f"Error: {result['error']}")

Search API fields:

Field Type Required Description
query string Yes Search query or question
paths list No Directories or files to search (e.g., ["/mnt/docs"]); falls back to SIRCHMUNK_SEARCH_PATHS if unset
mode string No "FAST" (default, greedy search 2-5s), "DEEP" (comprehensive analysis 10-30s), or "FILENAME_ONLY" (file discovery <1s)
enable_dir_scan bool No Enable directory scanning for file discovery in DEEP/FAST (default: true)
max_depth int No Maximum directory depth to search
top_k_files int No Number of top files to return
max_token_budget int No LLM token budget (DEEP mode)
return_context bool No Return full SearchContext with KnowledgeCluster and telemetry (default: false)

4. Manage the container

# View logs
docker logs -f sirchmunk

# Stop
docker stop sirchmunk

# Restart
docker start sirchmunk

# Remove container (data is preserved in the volume)
docker rm sirchmunk

# Remove data volume (caution: deletes all persisted data)
rm -rf /path/to/your_work_path