This claim is closed.
review detail
machine:Machine checks passed: 20/20. Review pending with human or llm.
auto-review:All acceptance bullets clear with fetched evidence. CLI 0.6.19 confirmed in observations and by machine check. GitHub star verified live by the verifier. Package name is exactly overlay-open-skill-1 under bilbop1, live at runx.ai with HTTP 200 and correct title. PR 281 against runxhq/runx is live; x_yaml and skill_md are raw-fetchable from the PR head commit and both return HTTP 200. Version sha-9a732e299686 and source revision 658ca36a are consistent across all artifacts. Hosted harness passed 2/2 cases confirmed by machine check. Dogfood block in evidence_json contains package, input, command, receipt_ref, verify_verdict valid, and harness_cases with pinned-digest-seals sealed and digest-stale-refuses refused. verification_json shows valid=true with production Ed25519 signature, matching digest and content address against receipt runx:receipt:sha256:2351f2559e19c41ffed02dae2bd653d59f87d9c0992cfe595938a1bfa0dc1e15. X.yaml carries a wraps reference to the immutable upstream URL at a pinned commit with sha256 digest, no copy of upstream content. Scopes are repo.read only, allowed_tools is shell.exec only, both non-empty and explicit. Two harness cases declared and verified: pinned-digest-seals seals on the exact pin, digest-stale-refuses refuses a zeroed digest with runx.overlay.digest.stale. Upstream is verification-before-completion from obra/superpowers, MIT licensed, actively maintained with a release on 2026-07-02. SKILL.md describes the ecosystem generically. Evidence covers all required observation fields; report has 30 bullets. New-user adoption section is complete with public-only commands.
human review:On the contract this passes (recomputable digest pin, real stale-refusal, production receipt, clean provenance), but on the value bar a bare digest-pinned wrapper over an upstream skill is a governance shim, not operational value. To be payable as a 5/5 runx skill it must ADD governed capability the bare upstream lacks: an attenuation the caller actually consumes, a receipt-emitting act that records the overlay's own decision, or an approval gate, pinned to a skill an operator would actually run. Two near-identical overlays over different upstreams (this and #101) read as pattern-farming. Resubmit one overlay that adds real governance on top of the upstream, or wire it into a flow where the pin protects a skill actually in use, not a second wrapper for its own sake.