--- name: solix-usage description: "Read usage across every linked provider — rate-limit windows, % left, reset times, walls, promo/credit pools, and measured run results. Use when asked about usage, quotas, rate limits, when a window resets, or where work can run." --- # Solix Usage Pull the full picture: ```text solix limits --history every provider window + 16-tick trend sparkline solix providers installed CLIs — cheapest model, cost tier, promos solix runs measured results: duration, outcome, ★ rating ``` `solix limits` prints one row per provider window the host has seen — account collectors read claude, codex, gemini, grok, copilot, cursor, kiro, and ollama directly; anything else reports through bot scrollback (ambient). Rows look like: ```text CLAUDE 5h open · 62% left · resets 4:30 PM ▂▄▅▆▅▄▃▂ CODEX weekly walled · resets Tue 9:14 AM GROK EXTRA open · 88% left ``` Reading it: - `walled` means the provider is rejecting work right now — route around it until `resets`. A wall with no named time yields as soon as the agent produces output again. - Window labels are the provider's own (`5h`, `weekly`, `session`). Non-base windows print uppercased — `PROMO`, `EXTRA`, `CREDITS` rows are bonus pools: spend them first on bounded tasks. - The sparkline is oldest→newest recorded % used. A rising tail means the window is draining faster than its reset will refill. - Missing `% left` means the provider didn't print a number — usable but unmeasured, not unlimited. Answer directly: provider, window, % left, reset. When the question is "where should this run", name the provider and why — open window first, promo pools for bounded work, never a walled one; `solix runs` shows what actually performed on this machine if a model recommendation is needed.