Research Papers

51 papers — methodology, data, and conclusions you can verify

research 18 min read intermediate

Endpoint Construction Is Product Construction

Why AI-native products depend on the capability boundary more than the chat surface. A practical model for treating endpoints as the product grammar that defines intent, schema, authority, state, limits, errors, evidence, and fallback.

Endpoint ConstructionMCPTool CallingAI-Native ProductAtlasPolicy OSWorkflow Trust LayerThree-Tier Framework
research 16 min read intermediate

The Loop-Operable Codebase

A CREATE SOMETHING field paper on converting loop-engineering hype into a bounded repo-native operating system using Linear, Symphony, worktrees, legibility contracts, policy artifacts, and explicit dispatch gates.

Loop EngineeringSymphonyLinearCodexHermesAgent HarnessWorktreesPolicy OSThree-Tier Framework
research 18 min read intermediate

The Proof Surface

Why agent work needs public receipts, private evidence, and owner authority once it leaves chat. A practical model for turning traces, deploys, approval notes, blocked states, delivery records, and workflow templates into business-readable proof.

Proof SurfaceReceiptsWorkflow Trust LayerPolicy OSAgent GovernanceDelivery RecordsOperator SurfaceThree-Tier Framework
research 17 min read intermediate

The Eval Evidence Layer

How Langfuse traces and Braintrust gates make agent workflows measurable. A quantitative model for turning Dify runtime traces, MCP eval gates, approval receipts, and blocked-state evidence into publish, hold, rollback, or graduation decisions.

Eval Evidence LayerLangfuseBraintrustDifyMCPPolicy OSAgent GovernanceObservabilityRelease Evidence
research 18 min read intermediate

The Policy OS Contract Bundle

Why governed AI workflows need a portable contract bundle before they need more autonomy. A practical model for defining MCP capability, agent behavior, outcome success, golden tasks, runbooks, and runtime graduation.

Policy OSContract BundleMCPDifyOpenAI Agents SDKGolden TasksAgent GovernanceSkills on MCPThree-Tier Framework
research 18 min read intermediate

The Workflow Trust Layer

Why agents need handoffs, approval states, and evidence before they need more tools. A practical operating model for mapping one workflow into auto-allowed, approval-needed, and blocked states before expanding autonomy.

Workflow Trust LayerPolicy OSMCPDifyOpenAI Agents SDKApproval StatesAgent GovernanceThree-Tier Framework
case-study 10 min read intermediate

Policy OS Applied to Development Infrastructure

Applying the Three-Tier Framework and Policy OS to the development workflow itself, demonstrating that agent governance emerges as a structural property at every scale.

Policy OSThree-Tier FrameworkPiAgent GovernanceQuality GatesMCPDevelopment Infrastructure
case-study 14 min read intermediate

Webflow Analyzer Productization

How CREATE SOMETHING translated reviewer-side analyzer infrastructure into creator-facing validation, autofill, screenshot packaging, and submission UX without collapsing trust boundaries.

WebflowAnalyzerProductizationCreator WorkflowReview SystemsThree-Tier FrameworkSubmission UX
case-study 16 min read intermediate

The Analyzer MCP: A Policy-Grounded Review Architecture

How CREATE SOMETHING turned Webflow template review into a multi-surface MCP system that joins Designer state, published-site evidence, policy ingestion, and governed review output.

Analyzer MCPWebflowMCPThree-Tier FrameworkPolicy as ArtifactReview SystemsObservability
research 22 min read intermediate

Composio in the MCP Delivery System

A decision-grade analysis of why Composio is included for commodity connectivity, how the wrap pattern protects brand and margin, and how delivery remains aligned to Database, Automation, and Judgment control boundaries.

ComposioMCPThree-Tier FrameworkWrap PatternAgent Outcome StackPolicy as Artifact
research 15 min read intermediate

Braintrust Trace Unsurfacing: Finding What Normal Aggregates Hide

How a 1,000-row trace snapshot exposed clustered permission failures, routing misses, and latent control-plane stalls.

BraintrustObservabilityMCPReliabilityExperiment DesignDashboarding
research 15 min read intermediate

The Wrap Pattern: Commodity Integration as Invisible Infrastructure

When MCP consumption is commoditized, the strategic response is to wrap — not build — the plumbing

A structural pattern for integrating commodity MCP vendors as invisible infrastructure while preserving the client-facing surface, the Intelligence Layer margin, and the Three-Tier alignment.

MCPWrap PatternCommodity IntegrationCreation MoatThree-Tier FrameworkInvisible InfrastructureAgent ArchitectureCloudflare WorkersModel Context Protocol