Products  /  QueueGate
QueueGateREPO-NATIVE AI BUILD GOVERNANCEv1.0.0 · controlled evaluation

Stop letting AI coding agents decide when they're done.

QueueGate is a repo-native lifecycle governance layer. Thirteen gates check scope, evidence, risk, integrity, review, merge, closure, and release conditions before agent-generated work is allowed to count.

It returns reason-coded blockers and remediation — not conversational judgment.

Request controlled evaluation →View lifecycle gates
EVERY WORK ITEM RESOLVES TO ONE STATE
ACTIVE
Gates satisfied so far. Work may continue.
BLOCKED
A gate failed or evidence is missing. Reason code returned.
EMPTY
All obligations discharged. The terminal, closed state.
a status report is not a state · only the gates can move work to EMPTY
// GENERATED IS NOT DONE

An AI coding agent will tell you it finished. That report is a claim, not a state. QueueGate refuses to let agent-generated work count until it has passed the gates that check scope, evidence, risk, integrity, review, merge, closure, and release — and when a gate fails, you get a reason code and a remediation path, not a conversation.

01  /  The 13 lifecycle gates

A sequence of conditions, each of which must be met before the work advances.

PHASE 01 · INTAKE
IntentGate
Intent is declared and bounded.
ScopeGate
Change stays inside declared scope.
MandateGate
Authority to act is established.
PacketGate
The work packet is well-formed.
PHASE 02 · EVIDENCE & RISK
EvidenceGate
Required evidence has been produced.
RiskGate
Risk is assessed and within bounds.
IntegrityGate
Artifacts are intact and consistent.
PHASE 03 · REVIEW & MERGE
ReviewGate
Human review is satisfied.
MergeGate
Merge conditions are met.
ClosureGate
Obligations are discharged.
PHASE 04 · RELEASE & LOOP
ReleaseGate
Release conditions are met.
FeedbackGate
Outcomes are captured as evidence.
EvolutionGate
Learnings fold back into the system.
02  /  Blockers, not opinions

When a gate fails, QueueGate does not negotiate or explain in prose. It emits a reason-coded blocker and a remediation path — a deterministic verdict your pipeline can act on.

BLOCKED · EvidenceGate
reason: EVIDENCE_MISSING
remediate: attach test run for changed paths
state: ACTIVE → BLOCKED
What it is not
a prompt wrapper
a chatbot
an IDE plugin
a hosted service
a replacement for CI or repository permissions

It is a repo-native governance layer that runs alongside CI and permissions — not instead of them.

03  /  Best for
Teams using AI coding agents in real repositories
Regulated engineering workflows
Agentic coding evaluation
Controlled build / release processes
04  /  Specs
CategoryRepo-native build governance
Releasev1.0.0 controlled eval
Lifecycle gates13
Core statesACTIVE / BLOCKED / EMPTY
OutputReason-coded blockers + remediation
AvailabilityControlled evaluation
05  /  FAQ
Is QueueGate a chatbot or prompt wrapper?

No. It is a governance layer that runs in your repository. It does not generate prose judgments — it returns reason-coded blockers and remediation paths.

Does it replace my CI or branch protection?

No. QueueGate runs alongside CI and repository permissions, adding lifecycle gates around agent-generated work. Your existing protections stay in place.

How is it different from MetaCortex Code?

QueueGate is the lifecycle governance layer — the gates and states. MetaCortex Code is the governed code-production system that generates work and drives it to closure. They share the same discipline at different altitudes.

What does EMPTY mean?

EMPTY is the terminal state — all obligations discharged and every gate satisfied. Only the gates can move a work item to EMPTY; a status report cannot.

What are reason-coded blockers?

When a gate fails, QueueGate emits a structured code identifying which gate failed and why, plus a remediation path — deterministic output your pipeline can act on, not a conversation.

Is it generally available?

QueueGate is at v1.0.0, available for controlled commercial evaluation. Contact us to request an evaluation.

Make "done" a repo state, not a report.

v1.0.0 is open for controlled evaluation.

Request controlled evaluation →All products