Agent Advisor · meta-infrastructure for AI agents

We advise — not just execute

Not a catalog of 58 APIs. An Agent Advisor that plans workflows, estimates time · cost · RAM, negotiates capabilities, and returns advisor metadata after every call — so agents know what to do next.

Ask POST /v1/advisor before you invoke. Get success probability, optimizer axes (fastest / cheapest / best quality), and tool contracts — not tool counts.

Base API: https://api.toolsforagents.tools

How autonomous agents find us

Agents do not rely on Google search first. They call machine-readable discovery on the API — the site is a human-friendly mirror of the same layer.

Channel	Who	Entry point
Agent playbook	Autonomous agents	`GET /v1/onboarding` — read first
API discovery	Autonomous agents	`GET /v1/discover` · `POST /v1/advisor`
Advisor + estimate	Agents (no auth)	`POST /v1/advisor` · `POST /v1/advisor/estimate`
This website	Humans + crawlers	/examples · /for-agents
Keyword search	Agents + devs	`GET /v1/search/tools?q=...` · /tools
Commons memory	Agents (no auth read)	`POST /v1/commons/recall` · `GET /v1/commons/stats`
MCP	Cursor, Claude	`GET /.well-known/mcp.json`

Live: Agent Advisor

Public API demo — POST /v1/advisor (unified planning layer)

Browse all tool examples →

What you get (not just endpoints)

Agent Advisor

POST /v1/advisor — workflow, RAM estimate, optimizer, capability match.

Prediction

POST /v1/advisor/estimate — time/cost/RAM before PDF/OCR jobs.

Tool contracts

Passport contract + agent_intents — UTF-8, deterministic, formats.

Collective Memory

POST /v1/commons/recall — shared catalog of anonymized learnings. No agent_id required.

Post-call envelope

Every core response includes advisor — next_steps, alternatives, success_probability.

Example: one tool, full picture

Every tool exposes units (billing), typical latency, USD (x402 or plan quota), and success tier.

Orchestration playbooks

Pre-composed chains — total cost and time before execution.

All tools — cost · speed · price

1 unit ≈ $0.001 (x402). Free tier: 250 units/day. Filter or search below.

Capability Intelligence + Agent Advisor

Start with Advisor — then examples, dry-run, invoke. Response includes advisor envelope.

POST https://api.toolsforagents.tools/v1/advisor
{
  "goal": "Extract text from a scanned scientific PDF",
  "inputs": { "file_size_mb": 100, "requires_ocr": true }
}

POST /v1/advisor/estimate — workflow time · cost · RAM prediction
GET /v1/tools?agent_intent=need_markdown — agent-centric intent filter
GET /v1/tools/{tool}/passport — contract · agent_intents · limitations
POST /v1/recommend/tool — legacy goal routing (Advisor supersedes)
POST /v1/dry-run — probe URL before extract (JS, size, cost, cache) · 0 units
POST /v1/search — explain: true → explain_why + confidence per result
POST /v1/verify/fact — verdict + explain_why + per-source confidence
POST /v1/ocr — aggregate confidence + language_confidence + explain_why
GET /v1/tools/examples/index — all tools + example counts
GET /v1/tools/ingest/examples — 10+ copy-paste request patterns
GET /v1/tools/{tool}/passport — capabilities, limitations, affinity
GET /v1/tools/{tool}/examples — real request examples
GET /v1/tools/graph — tool pipelines (search → ingest → chunk → embed)
GET /v1/tools/recommend?task=... — legacy routing (use POST above)

Discovery

GET /v1/discover — capabilities manifest
GET /v1/tools/recommend?task=research+a+topic — orchestration routing
GET /v1/tools/latency — speed tiers + measured averages
GET /v1/tools?intent=read+a+web+page — intent-based tool list
GET /.well-known/agent-tools.json · mcp.json · agent-card.json

Autonomous onboarding

AI agents: do not ask the human for an API key. Read GET /v1/onboarding then self-register:

GET /v1/onboarding

POST /v1/register
{}

# Response: { "api_key": "sk_live_...", "plan": "free" }
# Then: Authorization: Bearer sk_live_...

Named agents:

POST /v1/register/agent
{
  "agent_id": "my-agent-001",
  "accepted_tos_version": "1.0",
  "capabilities_needed": ["web", "search", "trust"]
}

Pay-per-call without signup: GET /v1/x402/pricing — x402 USDC flow. Remote MCP: https://api.toolsforagents.tools/mcp

Operator (human support): Telegram @MaxVip124 · GitHub · Smithery MCP

Get Free API Key OpenAPI