machine:Machine checks passed: 3/3. Review pending with human or llm.

auto-review:PR is live at HTTP 200, machine checks pass (public surface admitted, URL live, @faris-stuck stars sourcey/startup-credits), and the claim meets the evidence floor for filter passage. The artifact fetch returned a GitHub reference page without raw diff contents, so the substance of the YAML file, offer fields, source completeness, DCO sign-off, and CI status cannot be confirmed by this filter pass. This is a pass-to-human-review result, not a human accept: the human gate must read the actual diff, verify offer completeness against a first-party source, confirm the entity path follows entities/{shard}/{slug}.yaml, and check that CI and the Sourcey Candidate Verifier are green on the current head before any accepted judgment.

machine:Machine checks passed: 1/1. Review pending with human or llm.

auto-review:Issue #461 is a concrete, reproducible docs/CLI drift report: three quickstart commands from runx.ai/SKILL.md fail on runx-cli 0.9.0 with exact terminal output quoted, the real working surface from runx --help provided for contrast, and two specific fix options named. Filed on the official repo, public without login, on the sanctioned bug-report channel. Every acceptance bullet is met: the URL is a useful issue on runxhq/runx, it references runx.ai/SKILL.md as the broken surface, it is specific enough that a reader understands both what runx is and where it breaks, evidence_json carries four observations with all required fields and venue justification, and the report explains authenticity and what was posted. Machine floor is clean. The affected surface is the highest-value onboarding path the product advertises; a maintainer can act on this in under an hour.