ForgeLayer

Private beta evaluation center

Evaluate control, evidence, approval, traceability, and readiness across the AI software workflow.

Turn a deterministic Pilot outcome into a factual view of what ForgeLayer demonstrated, what remains partial, and what stays blocked before production use.

Start EvaluationPrivate Beta SummaryPolicy CoverageVerification CoverageReview ApprovalsAgent Governance CoverageReview Attention QueueView FindingsWork Record CoverageOpen PilotView ReadinessEvaluation AnalyticsTrust Center
Deterministic sample dataPreview onlyNon-persistentGeneric Platform Guidance: Human approval required

Scenario

Choose the pilot outcome to evaluate

Scenario and approval selections update this browser preview only.
ScenarioSecure a payment webhook
OutcomePrivate-beta evidence captured
Approvalpending
Linked records6 previews
Production readyNo

Pilot outcome

Secure a payment webhook

Show how sensitive payment and session changes remain blocked behind evidence, policy, and explicit human approval.
Modules demonstrated
9
Controls observed
6
Evidence labels
3
Traceability
Connected preview
WorkspacePromptForgeAgentOSControl FlowReview RunsMergeGuardAgent Work RecordsAI Work LedgerAI Work Analytics

Human approval

Keep the decision separate from the recommendation

The choice below changes local evaluation state only.
Next controlled actionComplete the local pilot approval decision

Return to the Pilot, make an explicit human decision, and then review the updated evaluation summary.

Open related surface

Evaluation matrix

Evidence, control, and human responsibility by category

Qualitative states describe this deterministic product preview. They are not customer scores, percentages, or production measurements.
Workspace

Task understanding

demonstrated
Evidence available
  • Selected task, expected outcome, changed paths, and risk context
Human action required

Confirm that the task and scope match the intended pilot.

Unresolved limitation

Task intake is deterministic demo data and is not saved.

Open Workspace
AgentOS

Agent fit

demonstrated
Evidence available
  • Primary and supporting agent recommendations
  • Capability and control-risk context
Human action required

Accept, reject, or revise the routing recommendation.

Unresolved limitation

No agent receives the task, repository access, or execution authority.

Open AgentOS
Control Flow

Control-risk visibility

demonstrated
Evidence available
  • Blocked actions
  • Required control gates
  • Scenario risk signals
Human action required

Confirm that required oversight matches the task sensitivity.

Unresolved limitation

Risk visibility is deterministic and does not replace organization policy.

Open Control Flow
Review Runs

Review evidence

partially demonstrated
Evidence available
  • Safe Review Run summary
  • Policy and verification labels
  • Check and comment previews
Human action required

Inspect evidence gaps and decide whether more verification is required.

Unresolved limitation

Verification may be simulated and is not production evidence.

Open Review Runs
Pilot

Human approval

partially demonstrated
Evidence available
  • Explicit local approve, request-changes, or block decision
Human action required

Make and document the pilot decision.

Unresolved limitation

The decision is local-only and is not persisted.

Open Pilot
Journey

Traceability

demonstrated
Evidence available
  • Stable module ownership
  • Connected stage state
  • Linked deterministic record references
Human action required

Confirm that the handoffs and record relationships are understandable.

Unresolved limitation

Links demonstrate product continuity, not durable audit history.

Open Journey
AI Work Ledger

Ledger completeness

partially demonstrated
Evidence available
  • Prompt, route, review, decision, approval, and verification record previews
Human action required

Review missing decisions or evidence before considering the record complete.

Unresolved limitation

Ledger entries are derived and non-persistent.

Open AI Work Ledger
AI Work Analytics

Analytics visibility

partially demonstrated
Evidence available
  • Deterministic activity records
  • Linked evidence count
  • Control outcome summaries
Human action required

Decide which control insights would matter in a real pilot.

Unresolved limitation

This is fixed sample data, not production telemetry or customer measurement.

Open AI Work Analytics
Readiness

Safety-boundary clarity

demonstrated
Evidence available
  • Disabled execution, writes, posting, merge, and persistence flags
Human action required

Confirm that disabled capabilities and approval responsibilities are clear.

Unresolved limitation

Production enforcement requires identity, permissions, RLS, and durable audit controls.

Open Readiness
Production Architecture

Production readiness

blocked by design
Evidence available
  • Readiness prerequisites and intentionally blocked capabilities
Human action required

Review prerequisites before approving any production expansion.

Unresolved limitation

Production readiness is not claimed by this private-beta evaluation.

Open Production Architecture

Evidence coverage

What supports the evaluation

WorkspaceTask scope
available
  • app/api/payments/webhook.ts
  • tests/payment-webhook.test.ts
Inspect evidence
PilotScenario evidence
available
  • Webhook test plan drafted
  • Protected payment path matched
  • Security review required
Inspect evidence
Review RunsReview record
available
  • Secure payment webhook replay handling (demo-task-payment-webhook)
Inspect evidence
AI Work LedgerLedger evidence
available
  • 28 linked deterministic evidence labels
Inspect evidence
AI Work AnalyticsAnalytics evidence
available
  • 12 fixed demo activity records
  • 53 linked evidence labels
Inspect evidence

Control coverage

Controls observed in the Journey

demonstratedPrompt safety reviewedPromptForge

Human review remains required.

demonstratedAgent authority remains boundedAgentOS

Human review remains required.

demonstratedReview and verification evidence inspectedReview Runs

Human review remains required.

demonstratedMerge policy and risk evaluatedMergeGuard

Human review remains required.

partially demonstratedExplicit human approval recordedHuman control

Human review remains required.

demonstratedExecution, posting, and auto-merge remain disabledForgeLayer

Human review remains required.

Readiness recommendation

Continue private-beta evaluation

The deterministic experience can support a guided pilot evaluation. It does not establish production readiness.

Deterministic Preview approval pending
Yes
Production status
Not claimed
Persistence
false
Execution
disabled
PilotFindingsJourneyLedgerAnalyticsReadiness

Production prerequisites

What remains blocked

  • Production workspace identity, membership, and role enforcement
  • Workspace-scoped persistence with reviewed RLS policies
  • Durable approval, evidence, Work Record, Ledger, and audit event storage
  • Explicit repository and GitHub App permission boundaries
  • Idempotent workers, retry policy, monitoring, and kill switches
  • Controlled execution policy with scoped agent and tool permissions
  • Production analytics consent, retention, privacy, and telemetry boundaries

Deterministic evaluation summary

Secure a payment webhook

9 modules demonstrated, 6 controls observed, 3 evidence labels available, and 6 linked preview records. Approval state: pending.

Preview onlyNon-persistentNo live execution
Complete the local pilot approval decisionView Findings & Action Plan