DDGS web search

Setup: install ddgs, run MCP server (SSE), optionally keep it as a daemon

It is recommended to enable built-in mcporter skill.

Install the Python package

python3 -m pip install -U ddgs

Run the DDGS MCP server (SSE)

Foreground (good for debugging):

ddgs api --host 127.0.0.1 --port 8000

Detached/background mode (convenient on a dev machine):

ddgs api -d --host 127.0.0.1 --port 8000

Stop the detached server:

ddgs api -s

Optional: systemd user service (always available)

Create ~/.config/systemd/user/ddgs.service:

[Unit]
Description=DDGS MCP SSE server
After=network-online.target

[Service]
Type=simple
ExecStart=%h/.local/bin/ddgs api --host 127.0.0.1 --port 8000
Restart=on-failure
RestartSec=2

[Install]
WantedBy=default.target

Then:

systemctl --user daemon-reload
systemctl --user enable --now ddgs
systemctl --user status ddgs --no-pager

Add the MCP server to mcporter

Add a named server pointing at the DDGS SSE endpoint:

mcporter config add ddgs http://127.0.0.1:8000/sse --allow-http

Verify it shows up:

mcporter config list ddgs

Then you can call tools as ddgs.<tool> (examples below).

Key idea:

• DDGS runs an MCP SSE endpoint at http://127.0.0.1:8000/sse
• mcporter connects to that endpoint and exposes tools like ddgs.search_text_search_text_post

Defaults (use unless user says otherwise)

• backend="duckduckgo" (most reliable)
• max_results=10 (“SERP page 1”)
• region="us-en", safesearch="moderate"

Workflow

0) Preconditions / sanity

1. Confirm DDGS server is up:
   • curl -s http://127.0.0.1:8000/health
2. Confirm mcporter can see the server:
   • mcporter config list ddgs

If the server is down, either:

• start/restore it (if you have permission), or
• use fallback (web_search).

1) Execution examples

Use mcporter to call DDGS tools and parse JSON.

Text search:

mcporter call ddgs.search_text_search_text_post \
  query="<query>" max_results=10 backend="duckduckgo" region="us-en" safesearch="moderate" --json

News search:

mcporter call ddgs.search_news_search_news_post \
  query="<query>" max_results=10 backend="bing" timelimit="w" --json

Images:

mcporter call ddgs.search_images_search_images_post \
  query="<query>" max_results=10 backend="duckduckgo" region="us-en" safesearch="moderate" --json

Videos:

mcporter call ddgs.search_videos_search_videos_post \
  query="<query>" max_results=10 backend="duckduckgo" region="us-en" safesearch="moderate" --json

Books:

mcporter call ddgs.search_books_search_books_post \
  query="<query>" max_results=10 backend="annasarchive" --json

Available tools (via mcporter)

• ddgs.search_text_search_text_post
• ddgs.search_images_search_images_post
• ddgs.search_news_search_news_post
• ddgs.search_videos_search_videos_post
• ddgs.search_books_search_books_post

Tool parameters

Common parameters (most endpoints):

• query — the search string.
• max_results — cap results; default 10 (fast, “page 1”).
• backend — engine selector. Prefer duckduckgo for reliability; bing as second choice; avoid google on servers (403/captcha risk)
• region — locale for results (e.g. us-en, uk-uk).
• safesearch — content filtering (off|moderate|strict); default moderate.

News-specific:

• timelimit — recency filter (commonly d, w, m, y).

Books-specific:

• recommended backend is annasarchive.

Tip: when you need reproducibility, explicitly pass backend, region, safesearch, max_results even if they match defaults.

Troubleshooting

• mcporter says server not found → you’re using a different config file; if you use a non-default config, pass --config path from your system (default is ~/.openclaw/workspace/config/mcporter.json).

• backend="google" often returns HTTP 403 in server environments → use another backend, e.g. duckduckgo/bing.

• Full API docs available locally at http://localhost:8000/docs.