Agent! is a 100% native Swift 6.2 / SwiftUI Mac app that runs an autonomous task loop against 18 LLM providers with your own API keys — reading a codebase, building an Xcode project, driving any app through the Accessibility API, or texting you the result over iMessage. Its native, bring-your-own-key design is the point for Mac users: work stays on the machine, and a task can't declare itself done until its goal criteria carry evidence.
That breadth is why the provider list matters: a fallback chain that auto-switches on 429s or timeouts is only as useful as the providers behind it. OrcaRouter speaks the same OpenAI chat-completions shape as the existing OpenRouter and Requesty entries, so it would slot into the picker as another route for long unattended runs.
I'm an engineer on the OrcaRouter team, and I'd like to propose OrcaRouter as an optional provider. It would not replace or change any existing provider — just add one more entry beside the current 18.
What it could add here
- Many models, one endpoint — chat, reasoning, and image models behind a single key, which fits the "One app. Any AI." premise of the picker.
- Automatic model routing and provider failover — a natural complement to
FallbackChainService, giving an overnight run more places to go after repeated 429s or timeouts.
- Prompt caching — the README already lists prompt caching as a reason to reach for Claude; OrcaRouter exposes it across a wider set of models.
- Budgets and usage tracking — pairs with the existing token accounting in
TokenUsageStore / LLMUsageView.
Integration
OrcaRouter exposes an OpenAI-compatible API and uses standard API-key authentication. That means the integration point is the one OpenRouter and Requesty already use: a new LLMProviderConfig with apiProtocol: .openAI plus a chatURL and modelsURL, registered in registerAllProviders() in Agent/Services/LLMProviderSetup.swift. Nothing is implemented or tested on my side — this issue is a proposal only.
Ecosystem
OrcaRouter is already used in open-source projects including RAGFlow, Dify, goose, and promptfoo.
Disclosure
OrcaRouter runs an optional open-source partner program: approved OSS projects can receive a 5% revenue share from usage attributed to their integration. Participation is entirely optional and is not a condition of any integration, and I'm glad to follow whatever disclosure or governance process this project requires.
Details are at https://www.orcarouter.ai/built-with. If this seems useful, I'd welcome your thoughts — and with your go-ahead I'd be happy to prepare an implementation PR.
Agent! is a 100% native Swift 6.2 / SwiftUI Mac app that runs an autonomous task loop against 18 LLM providers with your own API keys — reading a codebase, building an Xcode project, driving any app through the Accessibility API, or texting you the result over iMessage. Its native, bring-your-own-key design is the point for Mac users: work stays on the machine, and a task can't declare itself done until its goal criteria carry evidence.
That breadth is why the provider list matters: a fallback chain that auto-switches on 429s or timeouts is only as useful as the providers behind it. OrcaRouter speaks the same OpenAI chat-completions shape as the existing OpenRouter and Requesty entries, so it would slot into the picker as another route for long unattended runs.
I'm an engineer on the OrcaRouter team, and I'd like to propose OrcaRouter as an optional provider. It would not replace or change any existing provider — just add one more entry beside the current 18.
What it could add here
FallbackChainService, giving an overnight run more places to go after repeated 429s or timeouts.TokenUsageStore/LLMUsageView.Integration
OrcaRouter exposes an OpenAI-compatible API and uses standard API-key authentication. That means the integration point is the one OpenRouter and Requesty already use: a new
LLMProviderConfigwithapiProtocol: .openAIplus achatURLandmodelsURL, registered inregisterAllProviders()inAgent/Services/LLMProviderSetup.swift. Nothing is implemented or tested on my side — this issue is a proposal only.Ecosystem
OrcaRouter is already used in open-source projects including RAGFlow, Dify, goose, and promptfoo.
Disclosure
OrcaRouter runs an optional open-source partner program: approved OSS projects can receive a 5% revenue share from usage attributed to their integration. Participation is entirely optional and is not a condition of any integration, and I'm glad to follow whatever disclosure or governance process this project requires.
Details are at https://www.orcarouter.ai/built-with. If this seems useful, I'd welcome your thoughts — and with your go-ahead I'd be happy to prepare an implementation PR.