Ddg Search

ddg-search

Use the local skill script for lightweight DuckDuckGo search.

This skill exists to make fallback web search simple and low-noise instead of rebuilding ad hoc search helpers each time.

Quick rules

• Prefer your normal primary search tool when it is healthy.
• Use this skill when you need a fallback or a second source.
• Keep searches narrow and intentional.
• Prefer JSON output by default.
• Use --text only when human-readable terminal output is actually more useful.
• Use references/notes.md for quick guidance on when this tool fits.

Local setup

Prefer the skill-local script and cached virtualenv over any global CLI install. Treat the virtualenv under ~/.cache/ddg-search/venv as generated local state, not part of the skill itself. If the installer drops skill-local dotfiles, the bootstrap script recreates .gitignore automatically.

Bootstrap the local environment:

<skill-path>/scripts/bootstrap_venv.sh

After bootstrap, use:

<skill-path>/scripts/ddg-search

If the cached virtualenv is missing later, just run the bootstrap script again.

Commands

Show built-in help

<skill-path>/scripts/ddg-search help

Text search

<skill-path>/scripts/ddg-search search 'openclaw github'

News search

<skill-path>/scripts/ddg-search search 'bitcoin etf' --type news --timelimit d --max-results 10

Image search

<skill-path>/scripts/ddg-search search 'bitcoin logo' --type images --max-results 10

Video search

<skill-path>/scripts/ddg-search search 'openai launch' --type videos --timelimit w --max-results 10

Restrict to one site

<skill-path>/scripts/ddg-search search 'python dataclasses' --site docs.python.org

Reverse result order

<skill-path>/scripts/ddg-search search 'openai launch' --type videos --max-results 10 --reverse

Instant-answer lookup

<skill-path>/scripts/ddg-search instant 'weather berlin'

Resolve a DuckDuckGo bang

<skill-path>/scripts/ddg-search bang w 'OpenAI'

Workflow

1. Read references/notes.md if you want quick selection guidance.
2. Ensure the cached virtualenv is bootstrapped.
3. Use search for normal search, instant for instant-answer lookups, and bang when a DuckDuckGo bang is the cleanest path.
4. Parse JSON output by default.
5. Cross-check with another source when the question is important.

Expected outputs

The tool returns JSON by default. Parse it instead of scraping text.

Use --text when you want a readable terminal view.

• search returns query metadata plus results
• instant returns instant-answer style structured fields
• bang returns the DuckDuckGo URL plus the resolved redirect/final URL

Files

• Source repo: https://github.com/robin-liquidium/ddg-search-cli
• Launcher: scripts/ddg-search
• Python implementation: scripts/ddg_search.py
• Local bootstrap: scripts/bootstrap_venv.sh
• Notes: references/notes.md
• ~/.cache/ddg-search/venv is generated local state and can be recreated with <skill-path>/scripts/bootstrap_venv.sh

When to stop and ask

Stop and ask before:

• turning this into browser automation
• adding scraping-heavy flows
• adding provider-specific hacks that make the wrapper fragile
• changing default behavior from lightweight fallback search to something much broader