IBM Granite x402 Remote

The material states that no keys or environment variables are required, and there is no indication that the tool requests API keys, account passwords, or other highly sensitive credentials. The description mentions pay-per-use and USDC, but it does not state that the tool directly stores or reads users' wallet private keys.

It is known to access the remote endpoint three.ws; given its stated functions—remote AI chat, code, embeddings, and forecasting—user prompts or related request data will likely be sent to that service. The current material does not disclose more detailed data flows, retention policies, or third-party sharing.

System checks indicate that the tool has code-execution capability, which means it at least runs a local MCP process and handles tool invocations. This is a standard capability for this class of tool; the available material does not show requests for additional system privileges disproportionate to its stated purpose.

The material does not clearly state which local files, directories, or system resources it can read or write, so the data-access boundary is not fully transparent. Given that it is an executable MCP tool, it should be treated with least privilege by default, but there is currently no evidence of excessive data access beyond its stated scope.

The source is an official registry entry, with an open-source repository and updates within the last year, all of which are clear risk-reducing factors; no high-risk supply-chain red flags are evident. However, community adoption is very low (0 stars) and no license is declared, so auditability and reuse conditions still require further verification.