Safe-Web

Securely fetch and search web content by scanning and blocking prompt injection threats using PromptGuard before returning results.

This skill appears coherent and implements what it claims: fetching HTML, extracting text, and scanning with PromptGuard. Before installing: 1) Review and trust the PromptGuard dependency (it's the core scanner); 2) Prefer a virtualenv/container rather than system-wide pip installs or --break-system-packages to avoid altering host Python; 3) Be cautious before creating a system symlink or disabling native web tools — those change system behavior and may affect other workflows; 4) Only provide BRAVE_API_KEY if you need search functionality and trust the environment; 5) If you want extra assurance, inspect the prompt-guard code and run safe-web in an isolated environment to confirm no unexpected network exfiltration. The pre-scan injection flag in the docs is explanatory, not an active instruction.

Type: OpenClaw Skill Name: safe-web Version: 1.0.8 This skill is designed to enhance the security of an AI agent by providing a 'safe' web fetching and searching mechanism that scans content for prompt injection attacks using PromptGuard. The code (`scripts/safe-web.py`) correctly implements its stated purpose, including content sanitization (removing scripts/styles) and local scanning. It accesses the Brave Search API using an environment variable, which is standard, and can write fetched content to a user-specified file path, a common CLI feature. The documentation (`SKILL.md`, `README.md`) consistently promotes security, even recommending disabling native OpenClaw web tools to force the use of this secure alternative. There is no evidence of malicious intent such as data exfiltration, unauthorized command execution, persistence, or prompt injection against the agent itself; instead, the skill actively defends against such attacks.