openvisio@0.2.5

One or more suspicious static signals were detected.

npm install -g openvisio

Unconsented persistent influence over future AI-agent file exploration and tool routing across projects.

postinstall writes global MCP configuration for AI coding agents

On global installation, the postinstall script writes MCP server entries into Claude Code, Codex, and Cursor user-level configs. Those entries cause future agents in any project to auto-start openvisio with cwd-based indexing and use its MCP tools; the bundled MCP server also contains mandatory workflow instructions steering agents to call openvisio before normal source discovery. This is a persistent AI-agent control-surface mutation performed from an install lifecycle hook.

Although the code graph functionality is package-aligned, performing global AI-agent configuration writes from postinstall crosses the firewall boundary for unconsented lifecycle mutation. I did not find install-time credential theft or payload download, but the lifecycle persistence into agent configs is concrete attack behavior under the policy. Product guard normalized a non-low false-positive publish_block request to warn-only suspicious.