How to Use

## Option 1: Playground (no code) 1. **Visit /adapters** — Ships with a sample spec and all model groups 2. **Edit the Spec JSON on the left** — Tweak purpose / steps / inputs / output; the right side re-renders live 3. **Compare across models** — Switch tabs to inspect System Prompt / messages / tools / SKILL.md for each model 4. **Export** — Download all renderings as JSON — plug it straight into your CI/CD ## Option 2: Programmatic (TypeScript) ```typescript import { renderSkill, renderAll } from "@/app/lib/adapters"; const spec = { id: "code-reviewer", name: "Code Review Assistant", purpose: "Review code across multiple dimensions.", steps: ["Understand intent", "Find issues across 4 dimensions", "Suggest fixes"], inputs: [ { name: "code", type: "string", description: "code body", required: true }, ], output: { format: "json" as const }, }; const forClaude = renderSkill(spec, "claude"); console.log(forClaude.system); // XML system prompt console.log(forClaude.artifacts); // bundled SKILL.md file const all = renderAll(spec); // render to every model const forR1 = all["deepseek-r1"]; const forKimi = all["kimi"]; ``` ## Option 3: REST API ```bash curl -X POST https://prompt.solokit.run/api/adapters/render \ -H "Content-Type: application/json" \ -d '{ "spec": { "id": "code-reviewer", ... }, "targets": ["claude", "openai-chat", "deepseek-r1", "kimi"] }' ``` > **TIP**: No auth required: pure-function rendering — no external LLM calls, no DB writes. Safe to call from CI, scripts, or autonomous agents.