jiang-episode-quality-review — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited jiang-episode-quality-review (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Use this for source PRs that publish or revise one public episode or interview:
episode/<source-slug>
interview/<source-slug>This is an editorial QA gate. It does not rewrite the source, create lens pages, or process new videos. It decides whether the public read is good enough to merge, and if not, it gives concrete revision instructions.
content/lens/episodes/<source-slug>/read.jsoncontent/sources/videos/<source-slug>/transcripts/v1/transcript.clean.mdcontent/sources/videos/<source-slug>/transcripts/v1/transcript.clean.jsonlcontent/lens/evidence/videos/<source-slug>.semantic.json when present.phrase. Read enough surrounding transcript to judge whether the read carries the source's actual pressure.
and notable speaker/interviewer pressure.
existing lens pages, topic aliases, and prior signature moments. This catches reads that are technically valid but weaker than the project's known bar.
node ops/scripts/compile-content.mjs
node ops/scripts/validate-content.mjs
node ops/scripts/audit-episode-heat.mjs <source-slug> --strict --min 3
cd website && npm run buildPass only when the public page is worth reading without the transcript.
Required:
causal chains, provocations, and conceptual machinery.
answers. No invented FAQ questions.
answer.
Jiang's interpretation.
Fail when:
makes recognizable.
proposal tiers, or confidence buckets.
Post one of these to Moltnet and, when reviewing a PR, also comment on the PR.
Pass:
QA PASS <source-slug>: public read preserves the transcript's main pressure.
Checked: <files/commands>. Auto-merge <enabled/already enabled/not enabled because ...>.Needs work:
QA NEEDS WORK <source-slug>: <one-line reason>.
Fix:
1. <specific revision tied to transcript refs or section names>
2. <specific revision>
3. <specific revision>Do not write vague feedback. Every revision request should point to a concrete place in the read, transcript range, question list, marks, or source note.
Use strong judgment here. Aristotle is the guardrail that lets Virgil draft on a cheaper model: reject shallow or over-smoothed work, and ask Virgil to repair it instead of passing a page only because validators are clean.
If the PR passes and local validation passes, enable auto-merge:
gh pr merge <PR_NUMBER> --auto --squash --delete-branchIf the PR fails, do not enable auto-merge. Mention @virgil with the revision list and leave the PR open.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.