docs: document OrcaRouter as a named provider in the Crush guide

[putraperdana1207-pixel]

What

Adds OrcaRouter as a named provider example in the Crush guide.

Varlock's AI Tools docs show how to inject provider API keys into terminal coding agents so secrets never sit in a plain text .env or shell history. The Crush page already lists OPENROUTER_API_KEY and other provider keys, and shows an OPENROUTER_API_KEY entry in the .env.schema example. This PR adds the same first-class treatment for OrcaRouter: a short section showing the openai-compat provider entry in crush.json (mirroring how Crush documents custom OpenAI-compatible providers such as DeepSeek), an ORCAROUTER_API_KEY entry in the schema, and the varlock run -- crush launch command.

Why

OrcaRouter is an OpenAI-compatible AI gateway built for both models and agents. Like OpenRouter, it exposes a provider/model namespace across many models, but it also combines adaptive routing, automatic failover, zero-markup inference, observability, guardrails, and agent-tool governance behind the same endpoint. Adding it as a named provider in this guide means Crush users on varlock can use that stack directly, without treating OrcaRouter as an anonymous custom base URL.

It also runs gateway-level, zero-trust security for AI agents on the same endpoint, screening every prompt/response and governing every tool call on a default-deny basis, with no application code changes.

Verification

• bun run build:libs passed (27 packages).
• bun run astro build for @varlock/website passed, 130 pages built, and the generated /guides/ai-tools/crush/ page contains the new section.
• The OrcaRouter endpoint (https://api.orcarouter.ai/v1) returns 200 with a valid OpenAI-compatible response.

I'm an engineer on the OrcaRouter team.

[pullfrog]

Important

The configuration users are told to copy does not satisfy Crush's current model schema and should be updated before merging.

Reviewed changes in 639b0029: the new OrcaRouter setup in the Crush guide.

• Provider setup: Adds OrcaRouter's OpenAI-compatible endpoint and orcarouter/auto model configuration.
• Secret injection: Declares ORCAROUTER_API_KEY through a 1Password-backed env-spec value and launches Crush through varlock run.

  ^{｜ Fix all ➔ ｜ Fix 👍s ➔ ｜ View workflow run ｜ Using azure/gpt-5.6-sol ｜ 𝕏}

[pullfrog]

This explicit model entry is incomplete under Crush's current JSON schema: models also require pricing, context-window, token-limit, reasoning, and attachment fields, so users copying this example will not get a valid model configuration. Please switch the example to the current crushrc commands (model discovery can avoid hard-coding those fields), or supply a schema-complete legacy model.

Technical details

# Use a valid current Crush configuration

## Affected sites
- `packages/varlock-website/src/content/docs/guides/ai-tools/crush.mdx:27` - recommends the deprecated `~/.config/crush/crush.json` format
- `packages/varlock-website/src/content/docs/guides/ai-tools/crush.mdx:34` - starts an explicit model object that omits required fields

## Required outcome
- Give readers a configuration accepted by current Crush while retaining `$ORCAROUTER_API_KEY`, `https://api.orcarouter.ai/v1`, and access to `orcarouter/auto`.

## Suggested approach
- Use `~/.config/crush/crushrc` with `provider add orcarouter --type openai-compat --base-url "https://api.orcarouter.ai/v1" --api-key "$ORCAROUTER_API_KEY"`; current Crush discovers models for an `openai-compat` provider when no explicit model list is supplied.
- Upstream references: https://github.com/charmbracelet/crush/blob/main/docs/config/README.md and https://github.com/charmbracelet/crush/blob/main/schema.json