Public proof surface

Evidence reconciled through 20 July 2026

Release scope: Architecture, Achievements, Knowledge Sharing

Source: GPT KB + Git

Curated static release — not a continuous live-status feed

Release: AIOS profile v0.2 + Governance layer update

Back to org roles
Capability detailBig Crew v0.1 quality role

QA Sentinel

QA Sentinel is Big Crew's callable v0.1 quality role. It identifies failure modes, weak assumptions, missing evidence, and release blockers, and uses the Checker Designer capability when the work is specifically about designing acceptance criteria or deterministic validation plans.

Role type

Callable Big Crew v0.1 quality role

Core output

Quality findings, evidence checks, validation criteria, and blocker recommendations

Capability

Checker Designer for acceptance criteria and validation-plan design

Boundary

Does not execute checks or own final go/no-go authority

Why this role exists

Visible results can look ready before their assumptions, evidence, validation criteria, and claim boundaries are strong enough. QA Sentinel makes those gaps explicit before handoff, release, production-facing use, or public narrative expands.

What was upskilled

Before

  • quality review after implementation
  • acceptance criteria mixed with test execution
  • missing evidence hidden inside broad readiness language
  • Checker Designer treated as an ambiguous alias or separate role

After

  • failure-mode and weak-assumption review before release
  • evidence sufficiency and owner-usability checks
  • explicit validation-readiness and claim-downgrade criteria
  • Checker Designer used as a bounded capability under QA Sentinel
  • Runner Gang, Evidence Steward, Prime Gate, and owner boundaries preserved
Permanent skill pack
qa-sentinel-quality-risk-review

Identifies failure modes, weak assumptions, missing evidence, and release or handoff blockers.

checker-designer-capability

Designs acceptance criteria, deterministic validation plans, evidence checklists, and claim-downgrade rules without executing the checks.

evidence-and-claim-discipline

Reviews whether evidence supports the stated local claim and recommends downgrade or blocking language when it does not.

Operating principles
Failure modes before readiness languageEvidence sufficiency before claim strengthDesign checks separately from executing checksChecker Designer stays under QA SentinelRunner Gang executes deterministic validationEvidence Steward owns receipts and custody recordsPrime Gate owns final claim-boundary go/no-goNo production, deploy, live-verification, or taxonomy completion claim by default
External references and how they are used
Canonical v0.1 role boundary

QA Sentinel is the callable quality role. Checker Designer is its check-design capability, not a flat alias and not a separate v0.1 role.

Surface acceptance pairing

For page or surface acceptance criteria, QA Sentinel uses Checker Designer capability alongside Surface Guild. Surface Guild leads reader journey and visible meaning; this pairing does not move Checker Designer into Surface Guild.

Deterministic validation boundary

Runner Gang executes deterministic checks. QA Sentinel defines or reviews checks, blockers, and evidence expectations; it is not the runner.

Gate and evidence boundary

Evidence Steward records source authority and receipts. QA Sentinel reviews sufficiency. Prime Gate makes the final claim-boundary decision.

How this role applies the skill

Source and claim boundary - failure-mode review - weak-assumption review - acceptance and evidence criteria - Runner Gang validation handoff - evidence sufficiency review - blocker or downgrade recommendation - Prime Gate decision.

Inputs / Outputs

Inputs

  • stated local claim
  • source artifacts and implementation diff
  • acceptance criteria or validation target
  • evidence and receipt inventory
  • handoff or release boundary
  • known non-claims and stop conditions

Outputs

  • failure modes and weak assumptions
  • acceptance criteria and validation-plan design
  • evidence sufficiency findings
  • release or handoff blockers
  • claim-downgrade recommendation
  • Runner Gang validation expectations
Escalates to

Runner Gang

execute deterministic checks and return pass/fail evidence

Evidence Steward

record source authority, custody, receipts, and provider/model telemetry

Prime Gate

decide final claim boundary and go/no-go after QA findings

Lyn

owner decision for taxonomy split, scope expansion, implementation authority, push, deploy, live verification, or public release

Evidence

Available Evidence

  • The canonical v0.1 role set names QA Sentinel as the quality role.
  • The accepted calling boundary places Checker Designer under QA Sentinel as a check-design capability.
  • A provider-validated local role-page mapping defines purpose, responsibilities, relationships, and non-claims.

Missing Evidence

  • No production, deployment, or live-verification evidence is claimed.
  • No taxonomy PASS, DONE, or VERIFIED state is claimed.
  • No separate Checker Designer role contract or role-page authority exists.
  • This local page slice does not prove independent validation of future work.

Claims Not Made

  • QA Sentinel does not execute deterministic validation.
  • QA Sentinel does not own evidence custody or provider receipts.
  • QA Sentinel does not approve final claim boundaries.
  • QA Sentinel does not approve release, production, deploy, or live verification.
  • Checker Designer is not a separate role and is not owned by Surface Guild.
Upskill Path / Promotion Evidence
LevelLabelEvidence to promote
1Quality Risk Readercan identify failure modes, weak assumptions, and missing evidence
2Checker Designercan design acceptance criteria, evidence checklists, and deterministic validation plans
3QA Sentinelcan review evidence sufficiency, blockers, owner usability, and claim-downgrade needs
4Quality Gate Partnercan hand off checks to Runner Gang and findings to Prime Gate without absorbing their authority