smart-spawn-api by deeflect/smart-spawn
npx skills add https://github.com/deeflect/smart-spawn --skill smart-spawn-api为任何任务挑选最佳 AI 模型。调用 API,获取模型推荐,并使用它进行生成。
无需插件。 适用于任何 OpenClaw 实例或任何 HTTP 客户端。
sessions_spawn 中使用返回的模型 IDGET https://ss.deeflect.com/api/pick?task=build+a+react+dashboard&budget=medium
响应:
{
"data": {
"id": "anthropic/claude-opus-4.6",
"name": "Claude Opus 4.6",
"score": 86,
"pricing": { "prompt": 5, "completion": 25 },
"reason": "Best general model at medium budget ($0-5/M) — score: 86"
}
}
然后生成:
sessions_spawn(task="Build a React dashboard with auth", model="anthropic/claude-opus-4.6")
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
| 参数 | 必填 | 描述 |
|---|
task | 是 | 任务描述或类别:coding, reasoning, creative, vision, research, fast-cheap, general |
budget | 否 | low ($0-1/M), medium ($0-5/M, 默认), high ($2-20/M), any |
exclude | 否 | 要跳过的模型 ID,用逗号分隔 |
context | 否 | 用于路由增强的标签,如 vision,long-context |
GET https://ss.deeflect.com/api/recommend?task=coding&budget=low&count=3
返回来自不同提供商的多样化模型。用于集体/并行生成。
GET https://ss.deeflect.com/api/compare?models=anthropic/claude-opus-4.6,openai/gpt-5.2
并列显示分数、定价、能力。
GET https://ss.deeflect.com/api/models?category=coding&sort=score&limit=10
按 score、cost、efficiency 或任何类别名称排序。
POST https://ss.deeflect.com/api/decompose
{"task": "Build and deploy a SaaS app", "budget": "medium"}
返回顺序步骤以及每个步骤的最佳模型。
POST https://ss.deeflect.com/api/swarm
{"task": "Research competitors and build pitch deck", "budget": "low"}
返回一个带有分配模型的并行任务依赖图。
如果您没有使用插件,仍然可以从 API 获得模块化角色/系统风格的提示。
发现可用的区块 ID:
GET https://ss.deeflect.com/api/roles/blocks
组合一个角色增强的提示:
POST https://ss.deeflect.com/api/roles/compose
{
"task": "Build a SaaS onboarding flow",
"persona": "fullstack-engineer",
"stack": ["nextjs", "typescript", "postgres"],
"domain": "saas",
"format": "full-implementation",
"guardrails": ["code", "security", "production"]
}
将返回的 fullPrompt 用作 sessions_spawn 的 task 参数。
对于任何需要子代理的任务:
low 表示廉价/快速,medium 表示质量,high 表示最佳可用id 作为 sessions_spawn 中的 model 参数/api/decompose 或 /api/swarm 将其分解为子任务,并使用推荐的模型生成每个子任务sessions_spawn 时不指定模型(回退到默认值)any 并重试GET https://ss.deeflect.com/api/status
显示模型数量、数据新鲜度、来源健康状况。数据每 6 小时从 5 个基准来源刷新一次。
每周安装数
1
仓库
GitHub Stars
1
首次出现
1 天前
安全审计
安装于
amp1
cline1
openclaw1
opencode1
cursor1
kimi-cli1
Pick the best AI model for any task. Call the API, get a model recommendation, spawn with it.
No plugin required. Works with any OpenClaw instance or any HTTP client.
1. GET ss.deeflect.com/api/pick?task=<description>&budget=<tier>
2. Use the returned model ID in sessions_spawn
GET https://ss.deeflect.com/api/pick?task=build+a+react+dashboard&budget=medium
Response:
{
"data": {
"id": "anthropic/claude-opus-4.6",
"name": "Claude Opus 4.6",
"score": 86,
"pricing": { "prompt": 5, "completion": 25 },
"reason": "Best general model at medium budget ($0-5/M) — score: 86"
}
}
Then spawn:
sessions_spawn(task="Build a React dashboard with auth", model="anthropic/claude-opus-4.6")
| Param | Required | Description |
|---|---|---|
task | Yes | Task description or category: coding, reasoning, creative, vision, research, fast-cheap, general |
budget | No | low ($0-1/M), ($0-5/M, default), ($2-20/M), |
GET https://ss.deeflect.com/api/recommend?task=coding&budget=low&count=3
Returns diverse models from different providers. Use for collective/parallel spawning.
GET https://ss.deeflect.com/api/compare?models=anthropic/claude-opus-4.6,openai/gpt-5.2
Side-by-side scores, pricing, capabilities.
GET https://ss.deeflect.com/api/models?category=coding&sort=score&limit=10
Sort by score, cost, efficiency, or any category name.
POST https://ss.deeflect.com/api/decompose
{"task": "Build and deploy a SaaS app", "budget": "medium"}
Returns sequential steps with optimal model per step.
POST https://ss.deeflect.com/api/swarm
{"task": "Research competitors and build pitch deck", "budget": "low"}
Returns a dependency graph of parallel tasks with models assigned.
If you are not using the plugin, you can still get modular role/system-style prompting from the API.
GET https://ss.deeflect.com/api/roles/blocks
POST https://ss.deeflect.com/api/roles/compose
{
"task": "Build a SaaS onboarding flow",
"persona": "fullstack-engineer",
"stack": ["nextjs", "typescript", "postgres"],
"domain": "saas",
"format": "full-implementation",
"guardrails": ["code", "security", "production"]
}
3. Use returned fullPrompt as the task for sessions_spawn.
For any task that needs a sub-agent:
low for cheap/fast, medium for quality, high for best availableid as the model parameter in sessions_spawn/api/decompose or /api/swarm to break it into subtasks, spawn each with its recommended modelsessions_spawn without a model (falls back to default)any and retryGET https://ss.deeflect.com/api/status
Shows model count, data freshness, source health. Data refreshes every 6 hours from 5 benchmark sources.
Weekly Installs
1
Repository
GitHub Stars
1
First Seen
1 day ago
Security Audits
Gen Agent Trust HubPassSocketPassSnykWarn
Installed on
amp1
cline1
openclaw1
opencode1
cursor1
kimi-cli1
AI Elements:基于shadcn/ui的AI原生应用组件库,快速构建对话界面
60,400 周安装
mediumhighanyexclude | No | Comma-separated model IDs to skip |
context | No | Tags like vision,long-context for routing boost |