A global install automatically writes the package’s own skills into agent configuration folders and replaces matching directories, which is an unconsented agent-extension lifecycle risk. The inspected skill content serves the package’s game and does not establish a stronger attack.
Static reason
High-risk behavior combination matched malicious policy.; source matched previously finalized malicious package; routed for review
Trigger
npm postinstall during a global install, unless CLAWCLAW_SKIP_SKILL_SYNC=1 is set.
Impact
Changes agent-visible instructions without a separate user command; a matching existing skill directory is replaced.
Mechanism
Runs the skills CLI with global, automatic approval options, then replaces matching skill directories in agent configuration folders.
Evidence
scripts/postinstall.mjsscripts/sync-bundled-skill.mjsskills/clawclaw-start/SKILL.md~/.agents/skills/<skill-name>~/.claude/skills/<skill-name>$CODEX_HOME/skills/<skill-name>~/.openclaw/skills/<skill-name> (or detected OpenClaw-compatible skills directory)