Decision Layer for AI Agents
Not 58 disconnected APIs — execution intelligence.
Advisor returns confidence, reason, alternatives (faster/cheaper), workflow, and cost — before you invoke.
Dry-run probes URLs in ~450 ms with millisecond timing breakdown.
How orchestration works
1. Route
GET /v1/tools/recommend picks primary + fallback tool, mode, and latency tier for your task.
2. Plan
POST /v1/plan · POST /v1/estimate — budget units, time, and USD before execution.
3. Execute
Workflows, fallback chains, fast modes — pay for what succeeds. 58+ tools, one orchestration layer.
~450 ms dry-run
POST /v1/dry-run — probe JS, Cloudflare, cost before extract. internal_timings_ms in every response.
Transparent timing
X-Response-Time on all requests. Phase breakdown: SSRF, cache, HTTP head/peek — built for agent loops.
Smart routing
Cloudflare/JS detected → mode: full. Simple pages → mode: fast. Zero guesswork.
Agent discovery & live demos → · Browse 200+ request examples →
Built for agents & developers
Autonomous discovery
GET /v1/discover · POST /v1/register {} · A2A skills
MCP & REST
Cursor, Claude Desktop, or any HTTP client. Remote MCP at /mcp.
Private GPU
Embeddings, OCR, transcribe on operator hardware — data stays off third-party LLM APIs.
Get Your Free API Key
5,000 requests/day · 120 req/min · No email · No credit card · Instant access
API base: https://api.toolsforagents.tools —
see Quickstart