agentconcierge

通过 curl 调用 AgentConcierge 推荐接口，依据用户角色、痛点及可选的工具栈、预算、团队规模，从目录中检索返回 5 条按匹配度排序的 AI 智能体推荐，含名称、简介、价格、匹配分与匹配理由，并附错误处理与限流说明。

gomisterb

热度 36.33

最近更新 2026-05-11 17:46:09

分类：搜索依赖：代码执行 / 第三方 API应用场景：推荐 / 研究

安装技能

复制下面的话术发给你的 Agent，它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill： npx secure-skills-cli add agentconcierge

Metadata

name: agentconcierge
description: Find and recommend AI agents for any workflow using AgentConcierge. Use when someone asks which AI tools or agents would help them for their role, task, or pain point. Searches 9,800+ agents across all major marketplaces and returns 5 ranked recommendations with match scores and reasons.
version: 1.0.0
license: MIT-0
metadata: {"emoji":"🤖","openclaw":{"requires":{"bins":["curl"]},"install":[{"kind":"brew","name":"curl"}]}}

AgentConcierge

Discover the best AI agents for any workflow by querying AgentConcierge — a catalog of 9,800+ agents across OpenAI, Anthropic, n8n, HuggingFace, Zapier, and more. Returns 5 personalized recommendations ranked by match score.

When to Use

"What AI agents should I use for [task]?"
"Find me the best tools for [role] who needs help with [pain point]"
"I'm a [job title] — what agents would save me the most time?"
"Recommend AI tools with a [$X] budget using [tools]"
"What's the best agent for email, content, data analysis, scheduling, etc.?"

When NOT to Use

You already know the specific agent name and just want info (search the web instead)
Non-AI tools or general software outside the AI agent space
Requests unrelated to discovering or comparing AI agents

Required Context

Before calling the API, collect from the conversation:

Field	Required	Example
`role`	Yes	"Sales / Business Development"
`painPoint`	Yes	"Cold outreach and follow-up cadences"
`tools`	No	"HubSpot, Slack, Google Workspace"
`budget`	No	"$50" or "$200" or "$500+"
`teamSize`	No	"solo" or "small team"

If role or painPoint are missing, ask one concise follow-up before calling the API.

API Call

curl -s -X POST https://agentconcierge.io/api/recommend \
  -H "Content-Type: application/json" \
  -d '{
    "role": "Sales / Business Development",
    "painPoint": "Cold outreach and follow-up cadences",
    "tools": "HubSpot, Slack",
    "budget": "$50",
    "teamSize": "small team"
  }'

Response shape:

{
  "recommendations": [
    {
      "name": "Instantly AI",
      "tagline": "Automates cold email sequences at scale",
      "category": "automation",
      "pricingModel": "subscription",
      "pricePerMonth": 37,
      "tags": ["email", "outreach", "sales"],
      "url": "https://instantly.ai",
      "matchScore": 94,
      "matchReasons": [
        "Built for SDRs doing high-volume cold outreach",
        "Native HubSpot integration to sync contacts automatically"
      ]
    }
  ]
}

Presenting Results

Format each recommendation clearly:

🤖 Found 5 AI agents matched to your profile:

1. **Instantly AI** — 94% match
   Automates cold email sequences and follow-ups at scale.
   ✓ Built for SDRs doing high-volume cold outreach
   ✓ Native HubSpot integration to sync contacts automatically
   💰 $37/mo — https://instantly.ai

2. **Apollo.io** — 91% match
   ...

After presenting, offer:

"Want to save these to your stack at agentconcierge.io/dashboard?"

Error Handling

Status	Action
429	Tell user the API is rate-limited; retry in 60 seconds
400	Missing `role` or `painPoint` — ask user for the missing field
500/503	Suggest visiting https://agentconcierge.io/search directly

Notes

Rate limit: 10 requests per 60 seconds per IP
Budget $0 or "free only" returns only free/freemium agents
Same role + different pain point = different agent set
Full catalog browsing: https://agentconcierge.io/search
See references/api-schema.md for complete request/response schema