- What is pekral/ai-olympus used for in Laravel?
- AI Olympus is a configuration-driven package for defining AI agents, connecting model providers, and orchestrating multi-step AI workflows in Laravel projects. It is positioned as a unified layer over multiple AI providers so you can swap drivers without rewriting application code. In the 0.1.3 release it is a dev-only Composer package, meaning it adds no service provider, routes, migrations, or runtime dependency to your product.
- How do I install AI Olympus with Composer?
- Install it as a Composer package with composer require pekral/ai-olympus and then invoke its binary or Composer scripts as documented. The release notes describe one Composer command plus a binary invocation. Check the package README for the exact package name and command because the repository may include additional Composer scripts such as build/skill-check and shell-self-tests.
- Does AI Olympus add a service provider or runtime dependencies to my Laravel app?
- No. The 0.1.3 technical evaluation states that AI Olympus is a dev-only Composer package with zero product-runtime footprint: no service provider, no routes, no migrations, and no runtime dependency. That makes it safe to install in development and CI without affecting your production application's bootstrapping or database.
- Which Laravel and PHP versions does AI Olympus support?
- The provided release notes do not publish a definitive PHP/Laravel support matrix; the 0.1.2 assessment flagged that as an open question and 0.1.3 does not close it. Before adopting, inspect the package's composer.json for PHP and illuminate/support constraints and test against your project's Laravel version. If you need a guaranteed matrix, ask the maintainer or open an issue.
- How does AI Olympus handle multiple AI providers and model swapping?
- It is designed as a provider-agnostic layer: you register agents and model providers in configuration, then swap between hosted and local models behind a single consistent interface. The goal is to avoid vendor lock-in and raw API calls scattered through your code. The actual provider list and driver configuration keys should be confirmed in the package config files.
- Can agents call my existing Laravel services and actions?
- Yes, tool calling is a core feature: agents can invoke your existing Laravel services and actions safely, which lets workflows reuse application logic instead of duplicating it. The 0.1.3 release notes also describe review rules around app/Actions/** and duplicate or repeated Action sequences, so keep your Action classes focused and well-named. Review the tool registration config to control which actions an agent may call.
- Is AI Olympus suitable for CI pipelines or does it need a developer machine?
- Some workflows are CI-friendly, such as the merge gate and loader tests, and the package ships a hermetic fake-binary test harness. However, the interactive-testing feature now mutates the local environment: it installs dependencies, migrates a local database, rebuilds caches, builds assets, starts processes, and cleans up branches. The release notes say it targets local only and uses documented project commands, but you should confirm whether your CI runner can satisfy those prerequisites before enabling it.
- How do I fake AI responses so tests stay fast and deterministic?
- The package advertises testability as a first-class feature: you can fake AI responses so your test suite does not hit real provider APIs. The 0.1.3 notes also mention a hermetic fake-binary test harness for loaders and shell self-tests via composer shell-self-tests. Use these fakes in unit and integration tests, and reserve real provider calls for manual or staging checks.
- What are the risks of AI Olympus agents modifying my developer environment?
- The 0.1.3 interactive-testing readiness section describes agents installing dependencies, migrating a local database, refreshing caches, building assets, starting application processes, and deleting a branch they created. The documented guardrails are local-target only, project's own documented commands, no tracked-file edits, no discarding uncommitted changes, and git pull --ff-only. Even so, this is a new trust boundary with your machine; run it in a disposable checkout or container and review the scripts before use.
- Is AI Olympus production-ready, and what open issues should I watch?
- AI Olympus is a dev-only tool, not a production runtime library, so production readiness depends on your delivery architecture: agent workflow, CI, and the merge gate. The 0.1.3 evaluation notes unresolved items from 0.1.2, including the PHP/Laravel support matrix, --prune deletion semantics, and a regression test for the merge-gate trust boundary. Treat it as an evolving development-workflow package and pin versions while those questions remain open.