The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and process search results from the internet via a local SearXNG instance. This creates a surface where malicious content embedded in search results could attempt to influence the agent's behavior.
Ingestion points: Search data is fetched from http://localhost:8888/search as described in SKILL.md and references/agent-usage.md.
Boundary markers: The instructions do not define specific delimiters or "ignore embedded instructions" warnings for the agent when parsing search content.
Capability inventory: The skill environment provides access to curl, jq, podman, docker, and uvx for further operations.
Sanitization: The provided scripts and examples (such as the Nushell helper and curl/jq patterns) do not include sanitization steps to filter or escape untrusted web content before it enters the agent's context.
[EXTERNAL_DOWNLOADS]: The scripts/start-searxng utility downloads the official SearXNG container image from the well-known Docker Hub registry (docker.io/searxng/searxng:latest).
[COMMAND_EXECUTION]: The skill executes local commands to manage containers (podman or docker) and suggests the use of the uvx tool to execute the qypi package directly from the Python Package Index (PyPI) for package metadata searches.

searxng-search