Livt can be used with repository-aware coding agents such as GitHub Copilot, Claude Code, Codex, and Eccelerators fine-tuned Livt models. The practical workflow is simple: describe intent, provide Livt context, review the plan, run the Livt CLI, and publish only after review.
Working With Agents
A good agent task names the component goal, expected public API, packages to reuse, and the tests that should prove the result. Keep tasks small enough that the plan and the final package can be reviewed.
Use the Livt agent context.
Add a component that receives bytes from Uart, parses a simple command,
and exposes the command as a public field.
Reuse Livt.IO where possible.
Create tests for reset, one valid command, and an incomplete command.
Run livt validate and livt test after implementation.Livt Agent Context
For general coding agents, start with the official Livt agent context at github.com/eccelerators/livt-agent. It gives the agent the Livt syntax, package layout, CLI workflow, examples, and review expectations.
Add the context to the workspace instructions, agent memory, prompt context, or RAG setup depending on the tool. This keeps the agent aligned with Livt conventions instead of generic HDL-like code.
Fine-Tuned Livt Models
Eccelerators also provides Livt-specialized models that integrate into common agent tools and VS Code workflows. They are useful for teams that want Livt-aware assistance with lower token usage, lower operating cost, European data-center deployment, or on-premise deployment.
Prompt-to-Package With Labs
Eccelerators Labs turns a prompt into a reusable Livt package. The normal flow is:
Prompt -> Plan -> Develop -> Review -> PublishA completed lab can provide source files, tests, package metadata, command results, and downloadable package artefacts. When available, VHDL output can also be reviewed or integrated downstream.
Publishing and Reuse
Labs can publish reviewed packages to the package registry. Downstream projects then use the same package workflow as any other Livt dependency.
livt search MyGeneratedIp
livt add MyGeneratedIp@0.1.0
livt sync
Agents should prefer official packages before creating new infrastructure: Livt.IO for peripherals, Livt.Net for networking, Livt.Math for arithmetic helpers, and Livt.Crypto for cryptographic building blocks.
Labs Examples
The 8-point FFT lab demonstrates reuse of Livt.Math helpers such as Q15, ComplexQ15, and TrigQ15. The FIR HLS migration lab turns an HLS/C FIR shape into a tested Livt package with ProcessSample, SetCoefficient, and Reset. The SecureDataBlock lab uses Livt.Crypto.Aes.Ctr128 instead of implementing AES locally.
AI-Assisted Migration
For migration, treat existing C, HLS, HDL, or prototype code as reference behavior. Ask the agent for a component boundary and test strategy first, then create a small Livt package, run livt validate, livt build, and livt test, and review the result before publishing or replacing existing logic.
See Migration for the broader migration strategy around boundaries, wrappers, staged replacement, and behavior preservation.
Human Review
The developer still owns component boundaries, timing assumptions, context choices, package dependencies, test scenarios, security decisions, and final release approval. The agent helps with structured implementation work; the engineering decision remains human.
Summary
Use Livt with coding agents by giving them the right context, asking for small reviewable packages, reusing official libraries, and proving results with the Livt CLI. For hosted prompt-to-package work, use Eccelerators Labs and publish reviewed packages through the normal registry workflow.