Back to directory

Vizra ADK

A retired Laravel package for building autonomous AI agents with tools, sub-agent delegation and persistent memory; the final release still works but gets no further updates, and the project has moved to a separate evaluation tool instead.

Vizra ADK

Vizra ADK was a Laravel package for building autonomous AI agents that reason, use tools and keep persistent memory, integrated with a Laravel application's own syntax and conventions. It is retired as of August 2026, with its final release receiving no further fixes, security patches or issue responses, though that release still runs on PHP 8.2 and up and Laravel 11 through 13.

It supported OpenAI, Anthropic and Google Gemini models through the Prism PHP library, sub-agent delegation of tasks to specialized agents, an extensible tool system for reaching databases, APIs and other services, persistent memory across sessions, and workflows combining sequential, parallel, conditional and looping steps. Agents could be triggered conversationally, on a schedule, from a webhook, from an event, or as a queued job.

Its successor project, vizra/evals, is a separate agent-evaluation framework built on the official Laravel AI SDK, not a drop-in replacement; anyone who still needs to build agents is pointed instead to that official Laravel AI SDK, and the ADK itself remains available to fork under its MIT license.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.