byok forge

// launch-config generators for ai coding clis

self-hosted · client-side · no telemetry

Point your Copilot & Codex CLIs at any model.

Pick a provider and model, tweak the options, and copy a ready-to-paste launch config. Everything runs in your browser — your API keys are referenced as $ENV_VARS and never leave the page.

copilot-cli.html

copilot alias forge

Generate GitHub Copilot CLI launch aliases & functions that wire up COPILOT_* env vars for any BYOK provider — then copy & paste into your shell profile.

14 providers shell + powershell alias · function token limits

open generator →

codex-cli.html

codex config forge

Generate OpenAI Codex CLI BYOK profiles — a [model_providers] block plus a downloadable ~/.codex/<profile>.config.toml, with the matching codex -p launch command.

responses api profile.config.toml 10 providers downloadable

open generator →

litellm.html

litellm config forge

Build a multi-model litellm-config.yaml proxy that exposes /v1/responses for any chat-only provider (Mistral, DeepSeek, Anthropic, Gemini…) — then plug it into Codex or Copilot as a custom endpoint.

14 providers multi-model config.yaml codex hookup

open generator →

⚉

How it works. Each tool is a single, self-contained HTML file — no build, no server, no install. Open them locally or host them anywhere static. API keys are referenced by environment-variable name (e.g. $OLLAMA_API_KEY); the generated configs contain references, not secrets.