hydrate-opencode-modelslisted

# Hydrate OpenCode Model Parameters ## Security Gate (MANDATORY — do this first) opencode config files contain API keys and auth tokens. **You MUST NOT read any opencode config file until the user explicitly consents.** ### Step 0: Ask for trust decision Use the `question` tool to ask the user: > "I need to read your opencode config to see which providers and models you have. The config contains API keys and tokens. Do you trust me to read it?" Provide two options: 1. **Trust — read config directly** — I'll read your opencode.json/opencode.jsonc, identify all custom providers with missing/bare model definitions, and hydrate them. 2. **Don't trust — tell me model names** — You tell me which model names to look up, I'll fetch specs from Models.dev and return the config block for you to paste yourself. **If the user chooses "Don't trust":** - Ask which model names they want configured (e.g. "glm-5.1, kimi-k2.6"). - Follow Step 1–3 below to fetch specs from Models.dev. - **Do NOT read or edit any opencode config file.** Output the JSON snippet to chat for the user to paste themselves. - Stop here — do not proceed to Step 4. **If the user chooses "Trust":** - Read the opencode config file(s) to identify custom providers and their models. - Proceed with Step 1–4 below. ## Core Pattern Models.dev (`https://models.dev/api.json`) is a JSON catalog keyed by provider ID. Each provider has a `models` map with full metadata. For custom providers opencode does NOT auto-inherit