No concrete malicious chain was found. The package nevertheless performs default, behavior-bearing AI-agent hook setup during its scaffold command, so it meets the policy threshold for a warning rather than a block.
Static reason
One or more suspicious static signals were detected.
Trigger
User runs `eai init` without `--no-gofer`, then uses Claude Code in that project.
Impact
Changes agent behavior and retains derived conversation context within the initialized project.
Mechanism
Project-local Claude hook installation with transcript-derived memory persistence.
Evidence
package.jsondist/commands/init.jsdist/lib/gofer-installer.jsresources/gofer/hook-scripts/agent-stop.mjsresources/gofer/hook-scripts/user-prompt-submit.mjsresources/gofer/hook-scripts/post-tool-use.mjsdist/lib/update-check.js.claude/settings.json.specify/scripts/hooks/user-prompt-submit.mjs.specify/scripts/hooks/post-tool-use.mjs