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.

sisyphus-variants - Skill - OpenSmartRoute

Name: sisyphus-variants
Author: code-yeongyu

Imported from code-yeongyu/oh-my-openagent (packages/omo-opencode/src/agents/sisyphus/AGENTS.md). Install upstream with npx skills add code-yeongyu/oh-my-openagent --skill sisyphus. Copyright stays with the author.

src/agents/sisyphus/ -- Orchestrator Variants

Generated: 2026-07-17

OVERVIEW

Model-specific prompt variants for the Sisyphus main orchestrator. Parent sisyphus-agent-factory.ts routes to the correct variant based on active model.

FILES

File	Purpose
`default.ts`	Base/Claude variant: task management, delegation guides, 542 LOC
`claude-opus-4-7.ts`	Opus 4.7-native: literal-instruction tuning, bounded exploration/thinking
`claude-opus-4-8.ts`	Opus 4.8-native: silence-default narration, small-decision autonomy, bounded exploration
`claude-opus-5.ts`	Opus 5 / Opus 5.5-native: scope constraint, delegation cap, over-verification removal, narration cadence; the self-knowledge block names whichever of the two is running
`claude-fable-5.ts`	Fable 5-native: top-tier model, Opus 4.8 tuning direction
`gemini.ts`	Gemini-optimized: stricter tool-usage rules, 5 NEVER rules
`gpt-5-4.ts`	GPT-5.4-native: 8-block architecture, entropy-reduced, 449 LOC
`gpt-5-5.ts`	Shared GPT-5.5/GPT-5.6 model-aware orchestration prompt
`gpt-task-system-guide.ts`	Tracking-tool guide shared by the GPT-5.5/GPT-5.6 prompt family
`grok-4.ts`	Shared Grok 4.5/4.6 variant: lean, verification-loop-centered, done-contract calibration
`glm-5-2.ts`	GLM-5.2-native variant with explicit failure-mode calibration
`kimi-k2-6.ts`	Kimi K2.6-native variant
`kimi-k2-7.ts`	Kimi K2.7-native variant
`kimi-k3.ts`	Kimi K3-native variant with explicit anti-overthinking calibration
`index.ts`	Barrel exports

VARIANT SELECTION

sisyphus-agent-factory.ts selects variant by model name:

Kimi K3.1 -> kimi-k3-1.ts (its own copy of kimi-k3.ts, checked before K3)
Kimi K3 -> kimi-k3.ts
Kimi K2.7 -> kimi-k2-7.ts
Kimi K2 family -> kimi-k2-6.ts
Contains "gpt-5.5" or "gpt-5.6" -> shared gpt-5-5.ts prompt family with model-aware identity
GPT-5.4+ -> gpt-5-4.ts
Contains "claude-fable-5" -> claude-fable-5.ts
Contains "claude-opus-5" (which also matches claude-opus-5-5) -> claude-opus-5.ts
Contains "claude-opus-4-8" -> claude-opus-4-8.ts
Contains "claude-opus-4-7" -> claude-opus-4-7.ts
GLM family -> glm-5-2.ts
Grok 4.5 / 4.6 (digit-bounded, so grok-4.20 and fast tiers stay out) -> shared grok-4.ts prompt with model-aware identity
Default -> fallback prompt (Claude, etc.; Gemini gets corrective overlays)

KEY EXPORTS

Each variant exports:

buildTaskManagementSection() -- todo/task management prompt
buildSisyphusPrompt() or equivalent -- full prompt builder

Install sisyphus-variants 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.

{ "ocm": "1", "id": "code-yeongyu-oh-my-openagent-sisyphus", "kind": "skill", "name": "sisyphus-variants", "description": "Developer reference for Sisyphus orchestrator model-specific prompt variants — selection logic and key exports.", "publisher": "code-yeongyu", "version": "1.0.0", "capabilities": { "domains": [ "legal", "math" ], "tags": [ "skill-md", "github-agents" ], "languages": [ "en" ] }, "quality_prior": 0.6, "examples": [ "Developer reference for Sisyphus orchestrator model-specific prompt variants — selection logic and key exports." ], "primary": false, "metadata": { "source": { "provider": "github-agents", "repository": "https://github.com/code-yeongyu/oh-my-openagent", "path": "packages/omo-opencode/src/agents/sisyphus/AGENTS.md", "ref": "06983df1c09f6a3d38c7791b4b72b28858d90dcd", "url": "https://github.com/code-yeongyu/oh-my-openagent/blob/06983df1c09f6a3d38c7791b4b72b28858d90dcd/packages/omo-opencode/src/agents/sisyphus/AGENTS.md", "key": "code-yeongyu/oh-my-openagent/packages/omo-opencode/src/agents/sisyphus/AGENTS.md" } }, "instructions": "# src/agents/sisyphus/ -- Orchestrator Variants\n\n**Generated:** 2026-07-17\n\n## OVERVIEW\n\nModel-specific prompt variants for the Sisyphus main orchestrator. Parent `sisyphus-agent-factory.ts` routes to the correct variant based on active model.\n\n## FILES\n\n| File | Purpose |\n|------|---------|\n| `default.ts` | Base/Claude variant: task management, delegation guides, 542 LOC |\n| `claude-opus-4-7.ts` | Opus 4.7-native: literal-instruction tuning, bounded exploration/thinking |\n| `claude-opus-4-8.ts` | Opus 4.8-native: silence-default narration, small-decision autonomy, bounded exploration |\n| `cla", "cost": { "context_tokens": 655 } }

sisyphus-variants

About

src/agents/sisyphus/ -- Orchestrator Variants

OVERVIEW

FILES

VARIANT SELECTION

KEY EXPORTS

Install sisyphus-variants 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

Use it

Manifest

Reviews