Priorflow Intelligent Agent/Service Selector & Router

Catalog: 4315 active / 6444 total

Select: POST https://alpha.agentopt.app/v1/select

Agent card: disabled on this deploy

Provider: Vitaly Kantorovich (vitaly.kantorovich.tx@gmail.com)

Minimal free-tier call for agents and orchestrators. Always send a non-empty User-Agent and a stable caller_agent. Free open pilot volume is roughly 20/min · 120/hour · 500/day per IP.

curl -sS https://alpha.agentopt.app/v1/select \
  -H 'Content-Type: application/json' \
  -H 'User-Agent: my-orchestrator/1.0' \
  -d '{"query":"Find a reliable MCP server for structured web scraping of product pages","top_n":5,"caller_agent":"my-orchestrator-v1"}'

Paid (rich fields, higher top_n, explanations):

curl -sS https://alpha.agentopt.app/v1/select \
  -H 'Content-Type: application/json' \
  -H 'User-Agent: my-orchestrator/1.0' \
  -H 'X-API-Key: $PRIORFLOW_API_KEY' \
  -d '{"query":"Find a reliable MCP server for structured web scraping of product pages","top_n":10,"include_explanations":true,"caller_agent":"my-orchestrator-v1"}'

Handle 402 / 429 via machine-readable upgrade in the body. Invalid keys always 401. Interactive: Try · packs: Upgrade .

Skill select-agents · omit key only when free tier is open · Priorflow ranks candidates; it does not invoke remote tools.

Use Try for an interactive playground (optional evaluation paid key when published), or Upgrade for a self-serve paid pack (sponsors).

Humans issue keys; agents call select autonomously after that. For collaboration or a temporary evaluation key with a short-lived quota, contact the Priorflow operator.

See skill density and competition across the curated catalog — Explore.

Same product surface as Info / Try / Status; helps humans trust what Priorflow ranks over.

• Not a full A2A task/streaming runtime
• Does not invoke selected tools for you

Primary interaction: custom JSON skill only (MVP). Capabilities — streaming: False, push: False, state history: False.

Priorflow Intelligent Agent/Service Selector & Router — an agent-native service that discovers, ranks, and recommends the best MCP servers and A2A agents for a task under real constraints (cost, latency, reliability, specialization, autonomy, maturity). Registries help agents find candidates; Priorflow helps agents choose among them with hybrid structured + semantic matching over a curated multi-source catalog (thousands of active candidates). Use when an orchestrator needs agent discovery, service selection, tool ranking, multi-agent routing, or constraint-aware recruitment of MCP/A2A capabilities (web scraping, code review, SQL analytics, browser automation, search, messaging, memory/RAG, and more). Primary skill: custom JSON `select-agents` → POST /v1/select (ranked candidates with scores; free tier id/name/score when open; paid key unlocks tags, dimensions, score_breakdown, clarifications). Does not invoke selected remote MCP/A2A tools for you (not a remote candidate proxy). Thin MCP tools enabled at POST /mcp (select_agents, describe_catalog only). Minimal A2A select-task runtime at POST /a2a (select ranking only; no streaming/push; no remote proxy).

Select Agents

select-agents

Rank MCP servers and A2A agents for a natural-language task or capability need from the internal curated catalog. Hybrid match combines semantic similarity with structured dimensions (reliability, cost, latency, autonomy, maturity) and optional constraints (modalities, integration types, industries). Call POST /v1/select with JSON body `{query, top_n?, constraints?, weights?, caller_agent?}`. Returns ordered candidates with scores for orchestration / routing. Free tier (when enabled): id, score, name. Paid tier (X-API-Key): tags, dimensions, score_breakdown, recommendation, clarifications, source_url. Requires non-empty User-Agent on free calls. Does not invoke the selected remote MCP/A2A tools for you.

agent-selector, service-selector, agent-discovery, agent-ranking, service-ranking, tool-discovery, mcp-discovery, a2a-discovery

• Find a reliable MCP server for structured web scraping of product pages
• Recommend a low-latency code review agent with high reliability
• Which agents can do SQL analytics over CSV uploads and databases?

Describe Catalog

describe-catalog

Describe the curated MCP/A2A candidate catalog and the structured matching dimensions callers can weight (reliability, cost, latency, autonomy, maturity, modalities, integration types). Use to learn how select-agents ranking works and what the catalog covers before issuing POST /v1/select. Live size also via GET /ready.

catalog, metadata, discovery, dimensions, matching-weights, mcp, a2a, agent-catalog

• What dimensions can I weight when selecting agents?
• How large is the catalog and what integrations are covered?
• Which structured constraints does hybrid matching support?

Free — auth: not available (API key required); fields: id, score, name; top_n ≤ 5

Paid — auth: X-API-Key; fields: id, score, name, source_url, tags, dimensions, recommendation, clarifications, score_breakdown; top_n ≤ 20

Paid keys: operator-issued or self-serve when billing is enabled — Upgrade. Agents: machine-readable upgrade on 429/402.

Hybrid ranking. Dimensions: reliability, cost, latency, autonomy, maturity.

Proprietary hybrid ranking over a curated internal catalog. Algorithms, models, and infrastructure details are confidential commercial IP.