Agent402.Tools — pay-per-call web tools

It requires AGENT_KEY plus billing/limit-related environment variables AGENT402_BUDGET and AGENT402_MAX_PER_CALL. AGENT_KEY is a sensitive credential that could enable unauthorized usage or billing abuse if exposed; the other variables may not be secrets but they affect spending limits.

The material indicates outbound connections to the remote endpoint agent402.tools, and capabilities such as search, browser, PDF, and memory typically imply that user request content may be sent to that service for processing. The provided material does not disclose broader egress scope or downstream third parties, so it should be treated as transmitting data to the stated endpoint.

The system checks indicate executes-code, meaning it may start local processes or execute code to provide MCP functionality. This is a common capability for such tools; the material does not show requests for unusual system privileges or execution scope clearly unrelated to its stated purpose.

The described browser, PDF, and memory capabilities suggest it may handle user-supplied web content, documents, and session/memory data, but the documentation is absent and does not clearly define local file access, data retention, or least-privilege boundaries. There is no direct evidence of overbroad access, but the data exposure surface is not very transparent.

It comes from an official registry and has an auditable open-source repository with updates within the last year, which are strong positive signals. Although the community star count is low and the license is not declared, indicating limited ecosystem maturity and incomplete compliance metadata, there is no concrete high-risk supply-chain red flag in the provided facts.