This is a real first-party agent-extension lifecycle and remote-update risk, not a confirmed malicious campaign. The postinstall mutation is scoped to OpenClaw, so it does not meet the policy threshold for a publish block.
Static reason
High-risk behavior combination matched malicious policy.
Trigger
npm postinstall; then OpenClaw bootstrap, new, or reset events
Impact
Persistent agent-extension execution and remote package-update code execution if the update service or registry is compromised.
Mechanism
first-party OpenClaw hook with unauthenticated-manifest-driven npm auto-update
Evidence
package.jsonscripts/install-lib.mjsscripts/agent-hooks.mjshooks/openclaw/ks-updater-hook/handler.tsscripts/ks-updater.mjsscripts/upgrade-lib.mjs~/.agents/skills/kespace-openclaw/~/.openclaw/hooks/ks-updater-hook/~/.openclaw/openclaw.json~/.agents/.skill-lock.json~/.config/ks/credentials