The package has no install-time hook, and the global setup is documented and explicitly user-triggered. However, setup also broadens Claude Code’s global permission allowlist for lore commands and MCP tools, so this warrants a warning as an agent capability change.
Static reason
High-risk behavior combination matched malicious policy.; source matched previously finalized malicious package; routed for review
Trigger
The user runs `lore setup`.
Impact
Lore commands and MCP tools can run without individual permission prompts in Claude Code; the change is disclosed and user-triggered, but broadens global agent settings.
Mechanism
The command writes global MCP and instruction configuration, and adds Bash(lore:*) and mcp__lore__* to Claude Code’s allowlist.
Evidence
dist/cli.jsdist/global-setup.js~/.gemini/config/mcp_config.json~/.gemini/config/skills/lore/SKILL.md~/.claude.json~/.claude/CLAUDE.md~/.claude/settings.json~/.cursor/mcp.json~/.codex/AGENTS.md~/.ssh/lore_account_ed25519~/.ssh/lore_account_ed25519.pub