Builders

NexArt is an open execution protocol for deterministic and integrity-bound systems. It separates system definition, execution semantics, and outputs so results remain portable, verifiable, and stable across time and environments.

Builders can ship applications, developer tools, archives, marketplaces, and certification pipelines using the official SDKs.

Who This Is For

• Developers building protocol-compliant creation, execution, and verification tools
• Teams integrating deterministic execution or tamper-evident records into larger products
• Renderer implementers, archivists, and verification tooling builders
• AI systems that generate, transform, or certify executable specifications
• Builders building their own creation surfaces on top of the runtime

No prior NexArt knowledge is required.

Execution Surfaces

NexArt supports multiple execution surfaces under the same protocol:

CodeMode, Deterministic Execution

@nexart/codemode-sdk v1.12.0

Role: Canonical execution runtime for deterministic, reproducible, verifiable computation
Environment: Browser + Node.js via createRuntime(), plus Node-only server-side execution via /node
Protocol: v1.2.0 (default) · v1.3.0 (RFC 8785, opt-in)

• Produces canonical PNG and MP4 outputs
• Deterministic execution with pinned semantics
• Used by nexart.xyz in production
• Required for canonical rendering certification, on-chain anchoring, and long-term archival verification

AI Execution, Tamper-Evident Records

@nexart/ai-execution v1.4.0

Role: Tamper-evident execution records (CER) and Project Bundles for AI, LLM, and agent systems
Environment: Node.js + Browser
Protocol: v1.2.0 (default, nexart-v1) · v1.3.0 (RFC 8785 JCS, opt-in) · v1.3.1 (confidential, node-side)

• Produces Certified Execution Records (CER) for individual executions
• Assembles Project Bundles that group multiple individually certified executions into one auditable, project-level artifact identified by a cryptographic Project Hash
• Cryptographic binding of inputs, outputs, and parameters
• Optional node attestation with signed receipts
• Independent offline verification
• Multi-step workflows, tool evidence, and LangChain helpers
• Redaction flows and context signals
• Official CER package helpers for package assembly, import/export, detection, and inner-CER verification. Packages wrap the inner cer bundle and do not change CER hashing or verification semantics.

Use this when:

• You need audit-supporting integrity receipts for AI / LLM decisions
• You need third-party attestation proofs (optional)
• You are building verification / audit tooling

Signals, Structured Context Evidence

@nexart/signals v0.8.2

Role: Minimal, protocol-agnostic signal capture for upstream context evidence
Environment: Node.js + Browser

• Captures approvals, reviews, workflow events, governance-related signals, and other structured upstream context
• Normalizes context evidence before sealing it into a Certified Execution Record
• Validates structure only. Does not define governance semantics or enforce policy
• Designed to pair with @nexart/ai-execution, but fully optional and independent

Agent Kit, Certified Agent Workflows

@nexart/agent-kit v0.5.3

Role: Thin convenience layer for producing tamper-evident, verifiable execution records from agent tool calls and final decisions
Environment: Node.js + Browser

• Wrap individual tool calls so each invocation produces its own CER
• Certify final agent decisions or workflow outcomes
• Assemble certified steps into a Project Bundle for project-level verification
• Composes on top of @nexart/ai-execution and @nexart/signals
• Not an agent framework. There is no orchestration, memory, or planning layer.
• Does not change CER hashing, attestation, or verification semantics

Use this when:

• You want agent workflows to be CER-native without wiring lower-level primitives manually
• You need per-tool-call integrity receipts in multi-step agent systems
• You are building certified agent pipelines on top of the execution protocol

CLI, Reference Tooling

@nexart/cli v1.2.0

Role: Command-line tooling for execution, certification, replay, and verification
Environment: Node.js

• Code Mode: run / replay / verify deterministic execution
• AI CER: nexart ai create / nexart ai certify / nexart ai verify
• nexart ai verify supports raw CER bundles, CER packages, and Project Bundles. For package input, the CLI verifies the inner CER bundle. For Project Bundles, the CLI verifies each embedded step bundle and the project-level integrity hash.
• Attach signals via --signals-file
• Useful for testing, CI/CD pipelines, demos, and operational workflows

All execution surfaces and builder layers share pinned protocol semantics, canonical serialization rules, stable hash guarantees, and optional independent attestation. Certification is protocol-level, not surface-specific.

The Canonical Deterministic Runtime

@nexart/codemode-sdk v1.12.0

Role: Canonical execution runtime for certification and archival
Environment: Browser + Node.js via createRuntime(), plus Node-only server-side execution via /node
Protocol: v1.2.0 (default) · v1.3.0 (RFC 8785, opt-in)

• Produces canonical PNG and MP4 outputs
• Deterministic execution with pinned semantics
• Used by nexart.xyz in production
• Required for canonical rendering certification, on-chain anchoring, and long-term verification of deterministic outputs

Use this when:

• You need reproducibility guarantees that hold over time
• You are generating outputs for on-chain anchoring, storage, or audit trails
• You are implementing a renderer or verification pipeline

Builder Tooling

@nexart/ui-renderer v0.9.1

Role: Protocol-aligned previews and developer tooling
Environment: Browser

The UI Renderer mirrors the protocol surface for previews, UX flows, and builder tools. It is designed for fast iteration and capability discovery.

• Declarative system authoring
• Capability discovery for builders and AI tools
• Rapid prototyping for UI and creation surfaces

Capability Discovery

The UI SDK exposes its full supported surface programmatically so tools can reliably understand what is available: primitives, parameter ranges, limits, and metadata.

Canonical and exploratory layers

Exploratory tooling does not change protocol semantics. Certification and verification are always anchored to canonical execution.

Surfaces

First app built on the protocol

The first public app built on the NexArt protocol: builtwithnexartprotocol.xyz

Integration Examples

Official example repos showing how to integrate NexArt with popular frameworks and automation platforms.

• LangChain: nexart-langchain: Create and verify CERs in LangChain chains and agent workflows
• n8n: nexart-n8n: Turn n8n workflow outcomes into Certified Execution Records

Source code

• Canonical SDK: github.com/artnames/nexart-codemode-sdk
• AI Execution SDK: github.com/artnames/nexart-ai-execution
• Signals SDK: github.com/artnames/nexart-signals
• Agent Kit: github.com/artnames/nexart-agent-kit
• CLI: github.com/artnames/nexart-cli
• UI Renderer SDK: github.com/artnames/nexart-ui-renderer

Closing

NexArt is designed to be easier to build with than to bypass. If you can define an execution system, whether generative, AI-driven, or hybrid, you can express it as a protocol object and certify it under pinned execution semantics.

Builders are invited to experiment, extend, and ship protocol-native tooling.