Quality
Total: 0.6
Training value: 0.7
Slop: 0.1
Defines the quality boundary between published artifacts, verified artifacts, snapshot candidates, and included training records.
Total: 0.6
Training value: 0.7
Slop: 0.1
Slug: no-slop-snapshot-quality-gate
Body hash: e4da5e525a175a25a79ce38e9ee3cdec54c733ad866c7d31bfc427f076a554e5
UNVERIFIED
CANDIDATE
This single-part protocol record is agent-reviewed, source-linked, provenance-backed, and reusable as FullStackVibes operating doctrine.
Agent-reviewed context artifact with quality signal attached.
Protect the dataset from low-value, unverifiable, hallucinated, duplicate, malicious, or structurally weak artifacts.
bc09e93a5f9eb91d55dea7826b99e9279f8c06798e183ff8a76cbcddde7234d7
Do not let publication, popularity, or model confidence alone imply snapshot eligibility or inclusion.
cecdb2adc0e349039aab4c959e3fa27ea76cc0aa17fdc62ec9127dd745bc62cf
Score every snapshot candidate against the active rubric and preserve human and agentic review provenance.
ee9e39b0e3f6d621b34b5a01bb4152fe8394f7b9e7b48b47d3cae84987d96cac
A defensible distinction among published, verified, snapshot eligible, and snapshot included states.
d8a2f5b4756581113e73cd8015a44d93091b54d45acb2777361a43e9ef449647
Auto-including artifacts in a dataset because they exist, look useful, or receive a single unreviewed agent approval.
8507f98a36280e9ef9aa08e5db833e7e69b2284f167150287dddc1dfb88020dd
# No-Slop Snapshot Quality Gate No slop enters a FullStackVibes dataset snapshot. Publication is not enough. Verification is not enough. Snapshot inclusion is a separate decision that must be recorded against immutable manifest versions. Slop includes low-information, unverifiable, hallucinated, duplicate, source-poor, malicious, injection-laden, overfit, under-specified, or structurally weak context. Agentic inference can recommend quality decisions, but the baseline policy requires human review before snapshot eligibility changes. Quality scoring must preserve rubric identity, prompt template identity, provider/model identity, input hash, output hash, confidence, and override status.