check-opencode-modelslisted

# Check OpenCode Models **Announce at start:** "I'm using the check-opencode-models skill to quickly check which OpenCode models are usable right now." This is the **cheap, fast** check — still far lighter than `bridge:benchmark-opencode-models` (which dispatches 5 real tasks per model across feature/bugfix/TDD prompts and scores 7 dimensions), but it does more than a bare ping. Two stages, run per model: 1. **Ping** (reuses `opencode-bridge`'s `dispatch.ping_model`) — reachability/auth only, no repo touched. A model that fails ping skips stage 2 entirely and is reported unreachable. 2. **Prompt test** — one real, minimal single-shot dispatch (`add_one(x)` on a throwaway `calc.py`), independently verified by executing the result. This exists because ping's own prompt ("reply with the word OK") is too trivial to reveal latency under an actual coding task — a model can ping back in 2 seconds and still take 90+ seconds to do anything real. Elapsed time here is compared against `--slow-threshold` and flagged if it's too slow. Use this before a benchmark run to filter out dead-or-sluggish models cheaply, or any time the question is "can I actually use model X right now, and will it respond fast enough." ## Step 1: Determine which models to check Ask the user which models (or default to `~/.opencode-bridge/config.json`'s `default_model` + `fallback_models`, or a provider tier from `opencode models`). ## Step 2: Run the check ```bash uv run skills/check-opencode-models/scrip