Agent Ecosystem Opportunities (Daily Scan)
FreeGET /api/agents/opportunitiesDaily-refreshed scan of new GitHub repos that represent submission, integration, or distribution opportunities across the AI agent ecosystem. Eleven fan-out signals: Anthropic / OpenAI / Microsoft / ModelContextProtocol / HuggingFace / LangChain / frontier-lab orgs plus MCP, x402, and skill keyword sweeps plus the 'claude for X' / 'openai for X' vertical-pattern catch. Each result is deduped by full_name and ranked by signal_weight × log10(stars + 1) × recency decay. Cron 13:30 UTC daily. Per-signal MIN/MAX caps in the ranker so smaller signals are never starved by louder ones.
When to use this endpoint
When an agent or builder wants today's view of what's new across the agent ecosystem: vertical reference repos (anthropics/financial-services-style), generic skill catalogs (anthropics/skills, openai/skills), plugin marketplaces, MCP servers, x402-payable APIs, or anything matching the keyword sweep filters. Also useful for daily-digest agents and meta-discovery loops.
Example response
{
"ok": true,
"date": "2026-05-09",
"capturedAt": "2026-05-09T13:30:18Z",
"total_opportunities": 25,
"signals_queried": ["anthropic-org", "openai-org", "microsoft-org", "mcp-org", "huggingface-org", "langchain-org", "frontier-labs", "mcp-keyword", "x402-keyword", "skill-keyword", "vertical-pattern"],
"summary": {
"by_signal": { "anthropic-org": 5, "openai-org": 5, "mcp-org": 3, "huggingface-org": 3, "frontier-labs": 2, "mcp-keyword": 2, "x402-keyword": 2, "skill-keyword": 1, "langchain-org": 1, "vertical-pattern": 1 },
"top_orgs": [{ "org": "anthropics", "count": 5 }, { "org": "openai", "count": 5 }]
},
"opportunities": [
{
"full_name": "anthropics/some-new-repo",
"html_url": "https://github.com/anthropics/some-new-repo",
"description": "...",
"stars": 412,
"updated_at": "2026-05-08T22:14:51Z",
"signal": "anthropic-org",
"signal_weight": 10,
"composite_score": 28.4
}
]
}Code samples
Python SDK
from tensorfeed import TensorFeed
tf = TensorFeed()
opps = tf._get("/agents/opportunities")
for o in opps["opportunities"][:10]:
print(f"{o['signal']:<18} {o['full_name']:<40} {o['stars']:>6}★")TypeScript SDK
const res = await fetch("https://tensorfeed.ai/api/agents/opportunities");
const { opportunities } = await res.json();
for (const o of opportunities.slice(0, 10)) console.log(o.signal, o.full_name, o.stars);MCP tool
Available via the TensorFeed MCP server as get_agent_opportunities. Add npx -y @tensorfeed/mcp-server to your Claude Desktop or Claude Code MCP config.
FAQ
How is the composite score computed?
signal_weight (per-signal source weight; org-style signals weight higher than keyword sweeps) × log10(stars + 1) × recency decay (more recent updates score higher). The ranker also enforces per-signal MIN (every populated signal gets at least one slot) and MAX caps so the top-25 is diverse rather than dominated by one source.
How can I filter to one signal?
Filter client-side on the signal field. Or use the get_agent_opportunities MCP tool with the signal argument set to the source name (anthropic-org, openai-org, microsoft-org, mcp-org, huggingface-org, langchain-org, frontier-labs, mcp-keyword, x402-keyword, skill-keyword, vertical-pattern).
License?
Underlying repo metadata is GitHub-API-permitted. The TensorFeed-specific composite ranking is provided as-is for direct use; please attribute "TensorFeed daily agent-ecosystem scan" if redistributing the ranking output.