The source review found no install-time execution or concrete attack behavior. The similarity flags point to bundled Playwright code whose inspected agent configuration is command-triggered and whose other inspected excerpts cover test, helper, and UI functionality.
Static reason
High-risk behavior combination matched malicious policy.; source matched previously finalized malicious package; routed for review; previous stored version diff introduced dangerous source
Trigger
A user runs the Navarch CLI or explicitly invokes Playwright’s init-agents command; no npm install hook triggers these behaviors.
Impact
No malicious impact was established; the inspected behavior supports guarded coding-agent sessions and repository test tooling.
Mechanism
Navarch writes guard settings and configuration in a session work directory. The Playwright generator writes local agent and MCP configuration when explicitly selected.
Evidence
package.jsondist/worktree-guard.cjsbin/worktree-guard-hook.cjsdist/session.cjsdist/adapters/gemini.cjsnode_modules/playwright/lib/agents/generateAgents.jsnode_modules/playwright/lib/program.js