How long does the integration take?

Change one base URL in an OpenAI-compatible client, or install the Python or TypeScript SDK; the quickstart routes a first request in a few minutes.

Will this lock us into another vendor?

No. The router is Apache-2.0 open source, your provider keys stay yours, and the API is OpenAI-compatible, so leaving is a change of base URL.

What happens when a provider fails?

The router falls back to the next model your policy allows and records the fallback in the trace, so an outage costs a retry, not an incident.

ask-claude - Prompt - OpenSmartRoute

Name: ask-claude
Author: suchamoneypit

Prompt file imported from suchamoneypit/WARHOST (.cursor/commands/ask-claude.md). Copyright stays with the author.

/ask-claude

Hand one task to Claude Code through claude -p and show its report. The text after this command is a task name from .cursor/rules/cursor-to-claude.mdc (stuck_bug, pre_commit_review, plan_second_look, slop_review, implement_plan), optionally followed by details. With no task name, treat the text as a stuck_bug-style review question.

Look up that task's model, effort, mode, and max_turns in .cursor/rules/cursor-to-claude.mdc. A manual call ignores when, but still respects enabled and the per-chat limits.
Build a short prompt: the details given, the relevant paths, and the claim or question. For slop_review, the prompt is /slop-review. For pre_commit_review, it is /review-and-verify plus the changed file list. For implement_plan, it is /implement-plan <path>.
Run from the repository root with full_network permission:

sh scripts/ask_claude.sh --task --mode --model --effort --max-turns ""
Show the report verbatim, then one line with your own take on which findings matter. Do not apply fixes unless asked.

# /ask-claude Hand one task to Claude Code through `claude -p` and show its report. The text after this command is a task name from `.cursor/rules/cursor-to-claude.mdc` (`stuck_bug`, `pre_commit_review`, `plan_second_look`, `slop_review`, `implement_plan`), optionally followed by details. With no task name, treat the text as a `stuck_bug`-style review question. 1. Look up that task's `model`, `effort`, `mode`, and `max_turns` in `.cursor/rules/cursor-to-claude.mdc`. A manual call ignores `when`, but still respects `enabled` and the per-chat limits. 2. Build a short prompt: the details given, the relevant paths, and the claim or question. For `slop_review`, the prompt is `/slop-review`. For `pre_commit_review`, it is `/review-and-verify` plus the changed file list. For `implement_plan`, it is `/implement-plan <path>`. 3. Run from the repository root with `full_network` permission: sh scripts/ask_claude.sh --task <name> --mode <mode> --model <model> --effort <level> --max-turns <n> "<prompt>" 4. Show the report verbatim, then one line with your own take on which findings matter. Do not apply fixes unless asked.

Install ask-claude and it becomes one more candidate the router can pick - when it fits.

A listing is a routing target with a manifest: what it does, which domains it covers, what it costs and who publishes it. Once installed it sits beside your own models and tools, is scored like any other candidate for each request, and shows up in the trace when it wins. Ratings come from workspaces that installed it, one per account.

{ "prompt": "# /ask-claude\n\nHand one task to Claude Code through `claude -p` and show its report. The text after this command is a task name from `.cursor/rules/cursor-to-claude.mdc` (`stuck_bug`, `pre_commit_review`, `plan_second_look`, `slop_review`, `implement_plan`), optionally followed by details. With no task name, treat the text as a `stuck_bug`-style review question.\n\n1. Look up that task's `model`, `effort`, `mode`, and `max_turns` in `.cursor/rules/cursor-to-claude.mdc`. A manual call ignores `when`, but still respects `enabled` and the per-chat limits.\n2. Build a short prompt: the details given, the relevant paths, and the claim or question. For `slop_review`, the prompt is `/slop-review`. For `pre_commit_review`, it is `/review-and-verify` plus the changed file list. For `implement_plan`, it is `/implement-plan <path>`.\n3. Run from the repository root with `full_network` permission:\n\n sh scripts/ask_claude.sh --task <name> --mode <mode> --model <model> --effort <level> --max-turns <n> \"<prompt>\"\n\n4. Show the report verbatim, then one line with your own take on which findings matter. Do not apply fixes unless asked.", "variables": [], "notes": "Hand one task to Claude Code through `claude -p` and show its report. The text after this command is a task name from `.cursor/rules/cursor-to-claude.mdc` (`stuck_bug`, `pre_commit_review`, `plan_second_look`, `slop_review`, `implement_plan`), optionally followed by details. With no task name, treat the text as a `stuck_bug`-style review question.", "metadata": { "source": { "provider": "github-cursor-commands", "repository": "https://github.com/suchamoneypit/WARHOST", "path": ".cursor/commands/ask-claude.md", "ref": "56166ae90b43c4aaa77ad7fda3f39979ce1e69b4", "url": "https://github.com/suchamoneypit/WARHOST/blob/56166ae90b43c4aaa77ad7fda3f39979ce1e69b4/.cursor/commands/ask-claude.md", "key": "suchamoneypit/WARHOST/.cursor/commands/ask-claude.md" } } }

ask-claude

About

/ask-claude

Use it

Install ask-claude and it becomes one more candidate the router can pick - when it fits.

What it costs to adopt - in engineering time.

One base URL

No vendor lock-in

Failover built in

Manifest

Reviews