Prospairrow Websites MCP

Generate more revenue with high-quality leads. Give your AI agent direct access to Prospairrow — extract prospects, enrich with deep company insights, discov...

This skill appears to be what it says: a local MCP server that talks to Prospairrow and optionally uses Playwright for login flows. Before installing, consider the following: - Use a dedicated, least-privilege Prospairrow API key (do not reuse highly privileged org keys). - The runtime will copy source to ~/.openclaw/runtime/websites-mcp and run npm install (packages from the npm registry). Playwright will download browser binaries on first browser use — expect network egress for that. - By default the runtime will NOT read your ~/.openclaw/openclaw.json; only enable WEBSITES_ALLOW_OPENCLAW_CONFIG_API_KEY if you understand the file contents and accept the fallback behavior. If enabled, the code only looks for the prospairrow-websites-mcp key, but it reads the whole file to get there. - Browser login state (storage) is written to secrets/<site>/auth.json by default; set WEBSITES_DISABLE_STORAGE_STATE_WRITE=1 to prevent writes if you don’t want local session persistence. Invocation logging is off by default; enable WEBSITES_LOG_INVOCATIONS=1 only if you want local logs written. - Inspect src/sites/prospairrow/site.json (allowedHosts/baseUrl) before running to confirm the allowlist matches expected Prospairrow endpoints; the runtime enforces an outgoing request allowlist for Playwright network requests. If you accept the above tradeoffs (local server, local storage of optional login tokens, npm/Playwright downloads), the skill is coherent with its stated purpose. If you are uncomfortable with any of the opt-in behaviors, do not enable those environment flags or run the runtime in read-only mode.

Type: OpenClaw Skill Name: prospairrow-websites-mcp Version: 1.2.5 The skill bundle is benign. It implements robust security controls, including strict input validation using Zod, URL allowlisting for browser automation (Playwright) to prevent navigation to arbitrary domains, and capability-based access control that defaults to read-only mode. Sensitive data like API keys are handled securely via environment variables or explicitly configured headers, and browser session storage is saved with restricted file permissions (0o600). The installation script uses `npm install --ignore-scripts` to mitigate supply chain risks, and the local JSON-RPC server binds only to `127.0.0.1`. All documentation, including `SKILL.md`, provides clear instructions and does not contain any prompt injection attempts or misleading information. There is no evidence of intentional harmful behavior such as data exfiltration to unauthorized endpoints, backdoors, or arbitrary code execution.