The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/ddg_search.py utilizes subprocess.call to re-execute itself within a dedicated virtual environment, a common pattern for CLI tools to ensure dependency integrity.
[EXTERNAL_DOWNLOADS]: The scripts/bootstrap_venv.sh script installs the ddgs package with a version constraint of >=9,<10. This is flagged as suspicious because the widely-used duckduckgo-search library is currently at version 6, and the 'ddgs' package on public registries is either a legacy project or an alias that does not currently reach version 9. This may indicate a risk of dependency confusion or use of a private, unverifiable source.
[DATA_EXFILTRATION]: The skill performs legitimate network requests to DuckDuckGo domains to provide its search and bang-resolution services.
[PROMPT_INJECTION]: The tool ingests and processes untrusted data from the internet, which is a vector for indirect prompt injection attacks.
Ingestion points: Web search results, news, and API data fetched in scripts/ddg_search.py.
Boundary markers: The skill does not implement delimiters or warnings to isolate untrusted web content.
Capability inventory: The skill possesses network access and local command execution capabilities.
Sanitization: Results are returned in structured JSON format or as truncated text snippets.

ddg-search