fazmin/cve-intel is a native Rust tool that helps with vulnerability response. The latest release is v0.3.23, and the provided materials list no changes for this version. The repository description advertises 40 data sources and 41 tools, risk scores, and an MCP server for AI agents. The README states that it looks up CVEs, dependencies, IP addresses and more, and provides risk levels, whether an issue is being exploited, and remediation advice based on evidence from public security sources.
This tool handles vulnerability information locally and supports a CLI, a REST API, and an MCP server. MCP is the mechanism that lets AI agents call tools through a defined interface. According to the README, AI assistants run locally by default, and a cloud provider can be chosen if desired. The repository description and the README cite different counts of tools and sources: the README says 31 tools and 25 sources, while the code analysis included in the materials found 32 tool definitions. Any description of the feature scope therefore has to be read with that mismatch in mind.
The materials also describe an architecture where security information collection is processed locally with a small model built into llama.cpp. A vulnerability, IP, hash, or domain mentioned in the question is located first, the tools are run, and the results are handed to the model, which then produces explanations and further tool calls. When the model server cannot be reached, only the tool results are summarized without the model. Risk score calculation and caching, the source client, the configuration repository, and server functionality are implemented as separate pieces in the code. On the other hand, there are no test files and only 12 commits, so the published materials alone are not enough to judge implementation quality well. Local processing is the standout feature, but any adoption decision should take the limited verification evidence into account.