Security audit
Security checks for vulnerabilities and agentic risk
ClawScan marks this skill suspicious because a scanner found a release validation description paired with a request to publish packages; that request does not prove publication succeeded.
Review the skill’s publication behavior and confirm that requesting package and plugin publication fits your needs before installing. Treat the verdict as a caution: scanners did not confirm a completed publication, and some referenced paths were unresolved, leaving the evidence incomplete.
Although invoking full-release CI fits the description, the explicit validation_purpose=publish setting and publication selection enabling npm and plugin publishing introduce a consequential capability not disclosed by the validation-oriented description. The chunk shows a publication request; it does not establish that publication ultimately succeeds.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
the read-only OIDC preflight also uses `npm-publish` and requires that tag.
Artifact-only preflights keep their existing refs and have no environment.
- Never approve a `plugin-clawhub-release.yml` child by hand. It is
receipt-verified on the bot route and needs no approval. The
`Artifact not found` line for `openclaw-clawhub-recovery-approval-<run>-1`
is a non-fatal probe. A late human approval fails at `Revalidate trusted tooling identity`
with `parent state completed/failure is not allowed by authorization route`