Agent Discovery by cleanexpo/g-suite
npx skills add https://github.com/cleanexpo/g-suite --skill 'Agent Discovery'Agent Discovery 技能支持对智能体仓库、市场和研究平台进行结构化扫描,以发现新的 AI 智能体,并可能将其集成到 G-Pilot 中。
此技能遵循 Action-Ledger-Result 模式,并与以下服务集成:
agent_discovery主要发现函数 - 扫描多个智能源以查找智能体。
agent_discovery(userId, query, options) → DiscoveryResult
| 参数 | 类型 | 描述 |
|---|---|---|
userId | string | 用于计费和审计的用户 ID |
query |
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
| string |
| 搜索查询 |
options.sources | string[] | 要扫描的源(默认为前 3 个) |
options.limit | number | 每个源的最大结果数 |
options.focusAreas | string[] | 优先考虑的能力领域 |
options.minCompatibility | number | 最低兼容性分数(0-1) |
check_compatibility兼容性评估 - 对单个智能体档案进行深度分析。
check_compatibility(userId, dossier) → CompatibilityCheckResult
| 源 | API 类型 | 能力 |
|---|---|---|
github | REST API | 仓库元数据、星标数、许可证 |
huggingface | REST API | 模型信息、下载量、流水线标签 |
langchain_hub | AI-scanned | 提示词、链、智能体模板 |
crewai_flows | AI-scanned | 多智能体工作流 |
arxiv | AI-scanned | 研究论文、技术 |
gpt_store | AI-scanned | 自定义 GPT 智能体 |
vertex_ai | AI-scanned | 企业级智能体解决方案 |
aws_bedrock | AI-scanned | 基础模型智能体 |
azure_ai | AI-scanned | 认知服务智能体 |
{
"agentName": "example-agent",
"source": "github",
"sourceUrl": "https://github.com/example/agent",
"description": "What this agent does",
"compatibilityScore": 0.85,
"securityRating": "A",
"uniqueCapabilities": ["capability1", "capability2"],
"requiredDependencies": ["node", "typescript"],
"integrationEstimate": "1-2 days",
"fuelCostEstimate": "50-100 PTS",
"lastUpdated": "2026-01-19T00:00:00Z",
"stars": 1234,
"maintainer": "example-org",
"license": "MIT"
}
{
"success": true,
"query": "marketing automation agent",
"sourcesScanned": ["github", "huggingface", "langchain_hub"],
"dossiers": [...],
"duration": 2500
}
{
"success": true,
"agent": "example-agent",
"score": 0.8,
"checks": [
{ "name": "Language Compatibility", "passed": true },
{ "name": "License Check", "passed": true },
{ "name": "Security Rating", "passed": true }
],
"recommendation": "integrate"
}
import { agent_discovery, check_compatibility } from '@/tools/agentDiscoverySkill';
// 为特定能力发现智能体
const result = await agent_discovery(
'user_123',
'multi-agent orchestration',
{
sources: ['github', 'langchain_hub'],
limit: 10,
focusAreas: ['automation', 'workflow'],
minCompatibility: 0.7
}
);
// 深度检查特定候选者
for (const dossier of result.dossiers.slice(0, 3)) {
const check = await check_compatibility('user_123', dossier);
if (check.recommendation === 'integrate') {
console.log(`Recommended: ${dossier.agentName}`);
}
}
此技能是绑定到 Agent Scout 智能体的主要工具:
Agent Scout
├── agent_discovery (primary)
├── check_compatibility
├── web_intel
├── structured_scraper
└── deep_lookup
| 评级 | 描述 | 标准 |
|---|---|---|
| A | 可信 | 流行、维护良好、宽松许可证 |
| B | 安全 | 已知维护者、宽松许可证 |
| C | 需审查 | 信息有限,需要人工审查 |
| D | 需谨慎 | 未知维护者、限制性许可证 |
| F | 拒绝 | 存在安全问题、已弃用 |
| 操作 | 燃料成本 |
|---|---|
| 快速扫描(3 个源) | 25 PTS |
| 深度扫描(所有源) | 100 PTS |
| 单个兼容性检查 | 10 PTS |
每周安装数
0
仓库
GitHub 星标数
1
首次出现时间
Jan 1, 1970
安全审计
The Agent Discovery skill enables structured scanning of agent repositories, marketplaces, and research platforms to discover new AI agents for potential integration into G-Pilot.
This skill follows the Action-Ledger-Result pattern and integrates with:
agent_discoveryPrimary discovery function - Scans multiple intelligence sources for agents.
agent_discovery(userId, query, options) → DiscoveryResult
| Parameter | Type | Description |
|---|---|---|
userId | string | User ID for billing and auditing |
query | string | Search query |
options.sources | string[] | Sources to scan (defaults to top 3) |
options.limit | number | Max results per source |
options.focusAreas | string[] | Capability areas to prioritize |
options.minCompatibility | number | Minimum compatibility score (0-1) |
check_compatibilityCompatibility assessment - Deep analysis of a single agent dossier.
check_compatibility(userId, dossier) → CompatibilityCheckResult
| Source | API Type | Capabilities |
|---|---|---|
github | REST API | Repository metadata, stars, license |
huggingface | REST API | Model info, downloads, pipeline tags |
langchain_hub | AI-scanned | Prompts, chains, agent templates |
crewai_flows | AI-scanned | Multi-agent workflows |
arxiv | AI-scanned |
{
"agentName": "example-agent",
"source": "github",
"sourceUrl": "https://github.com/example/agent",
"description": "What this agent does",
"compatibilityScore": 0.85,
"securityRating": "A",
"uniqueCapabilities": ["capability1", "capability2"],
"requiredDependencies": ["node", "typescript"],
"integrationEstimate": "1-2 days",
"fuelCostEstimate": "50-100 PTS",
"lastUpdated": "2026-01-19T00:00:00Z",
"stars": 1234,
"maintainer": "example-org",
"license": "MIT"
}
{
"success": true,
"query": "marketing automation agent",
"sourcesScanned": ["github", "huggingface", "langchain_hub"],
"dossiers": [...],
"duration": 2500
}
{
"success": true,
"agent": "example-agent",
"score": 0.8,
"checks": [
{ "name": "Language Compatibility", "passed": true },
{ "name": "License Check", "passed": true },
{ "name": "Security Rating", "passed": true }
],
"recommendation": "integrate"
}
import { agent_discovery, check_compatibility } from '@/tools/agentDiscoverySkill';
// Discover agents for a specific capability
const result = await agent_discovery(
'user_123',
'multi-agent orchestration',
{
sources: ['github', 'langchain_hub'],
limit: 10,
focusAreas: ['automation', 'workflow'],
minCompatibility: 0.7
}
);
// Deep check a specific candidate
for (const dossier of result.dossiers.slice(0, 3)) {
const check = await check_compatibility('user_123', dossier);
if (check.recommendation === 'integrate') {
console.log(`Recommended: ${dossier.agentName}`);
}
}
This skill is the primary tool bound to the Agent Scout agent:
Agent Scout
├── agent_discovery (primary)
├── check_compatibility
├── web_intel
├── structured_scraper
└── deep_lookup
| Rating | Description | Criteria |
|---|---|---|
| A | Trusted | Popular, well-maintained, permissive license |
| B | Safe | Known maintainer, permissive license |
| C | Review | Limited info, needs manual review |
| D | Caution | Unknown maintainer, restrictive license |
| F | Reject | Security concerns, deprecated |
| Operation | Fuel Cost |
|---|---|
| Quick Scan (3 sources) | 25 PTS |
| Deep Scan (all sources) | 100 PTS |
| Individual Compatibility Check | 10 PTS |
Weekly Installs
0
Repository
GitHub Stars
1
First Seen
Jan 1, 1970
Security Audits
agent-browser 浏览器自动化工具 - Vercel Labs 命令行网页操作与测试
140,500 周安装
| Research papers, techniques |
gpt_store | AI-scanned | Custom GPT agents |
vertex_ai | AI-scanned | Enterprise agent solutions |
aws_bedrock | AI-scanned | Foundation model agents |
azure_ai | AI-scanned | Cognitive service agents |