The install hook only prints an explicit setup suggestion, and agent integration changes are exposed through user-invoked CLI commands. Inspected model downloads and store pruning are consistent with the package’s stated code-search function; no concrete attack behavior was found.
Static reason
One or more suspicious static signals were detected.; source matched previously finalized malicious package; routed for review; previous stored version diff introduced dangerous source
Trigger
Package installation prints a setup message; plugin changes require an explicit CLI command.
Impact
The inspected behavior does not establish unauthorized configuration changes, credential exposure, or destructive activity.
Mechanism
The postinstall script prints instructions, while separate Claude and Codex commands perform plugin setup. The cleanup helper uses a verified store lease to prune table history.
Evidence
package.jsonscripts/postinstall.jsdist/commands/claude-code.jsdist/commands/codex.jsdist/lib/store/lance-cleanup.jslance-maintenance/prune.pydist/lib/workers/download-worker.js