Augmented Search

Provides concurrent web search and code search capabilities for Agents with hybrid retrieval. Supports searching multiple keywords simultaneously, Embedding...

This skill appears to be a legitimate augmented-search integration, but exercise caution before installing: 1) The registry metadata omitted the required SEARXNG_URL — ensure you set SEARXNG_URL to a SearXNG instance you control (avoid public instances if you care about query privacy). 2) Optional embedding and Context7 API keys are reasonable for hybrid retrieval, but only provide keys you trust. 3) The docs suggest installing Ollama using curl | sh — avoid piping unknown install scripts; prefer vetted installation methods or inspect the script first. 4) The service runs as a Docker or npm application and will transmit queries to configured endpoints (SearXNG, embedding services) — review and verify those endpoints and container images (ghcr.io) before use. 5) Because the registry metadata and SKILL.md disagree, verify configuration expectations in the GitHub repo and test in an isolated environment first.

Type: OpenClaw Skill Name: performing-searches Version: 1.1.3 The skill bundle is classified as suspicious primarily due to the inclusion of a `curl | sh` command in `reference/installation.md` for installing Ollama, which is an inherently high-risk method for executing remote code, even if for a legitimate prerequisite. Additionally, the skill defines tools like `read` that accept arbitrary `urls` and `search` that accepts `searchedKeywords` as input. While the skill itself does not instruct the agent to exploit these, such parameters present a significant vulnerability surface (e.g., SSRF, shell injection) if the underlying `augmented-search` service is not robustly secured against malicious input, which an AI agent could potentially be prompted to exploit.