QVeris for AI Agents — Frequently Asked Questions

QVeris is a capability routing network that lets AI agents discover, inspect, and call 10,000+ real-world, verified capabilities across 15+ categories through a single API key and a unified Discover → Inspect → Call protocol. Each call returns a pre-settlement billing breakdown, with final settlement auditable via usage_history and credits_ledger.

Agents send a natural-language query to the Discover endpoint, receive ranked capability matches with quality signals, then Inspect the schema and Call the selected capability. Discover and Inspect are free; Call follows each capability's billing rule with pre-settlement and final-settlement audit surfaces.

What is QVeris and how does it help AI agents?: QVeris is a capability routing network for AI agents. Instead of hard-coding individual API integrations, agents use QVeris to discover, inspect, and call 10,000+ real-world, verified capabilities across finance, vision, documents, media, and more — all through one API key and one unified protocol.
How does the Discover, Inspect, and Call protocol work?: Discover accepts a natural-language query and returns ranked capability matches with quality signals. Inspect retrieves the full schema, parameters, and the capability's billing_rule. Call executes the capability and returns structured output plus a pre-settlement billing breakdown when available. Discover and Inspect are free; Call consumes pay-as-you-go credits.
How do I check whether a Call was actually charged?: Use usage_history with the execution_id to inspect charge_outcome (charged / included / failed_not_charged / failed_charged_review) and the actual_amount_credits. Use credits_ledger to see balance movements. Both endpoints support context-safe modes: summary, search with precise filters, or export_file for large analysis.
What kinds of capabilities can agents access through QVeris?: QVeris routes to capabilities in 15+ categories including real-time market data, media intelligence, multimodal vision (OCR, image recognition), document parsing, knowledge retrieval, live search, and signal analysis. New providers are added continuously.
How is QVeris different from calling APIs directly?: Direct API integration works for a small number of fixed endpoints. QVeris adds a routing layer that handles capability discovery, quality scoring, provider selection, and unified billing with pre-settlement transparency — so agents can reach many providers without managing individual API keys, rate limits, or contracts.
How do I integrate QVeris into my agent?: The recommended method is the QVeris CLI (v0.5.0) — it runs as a subprocess with zero schema tokens consumed (tool registries never enter the LLM context). Alternatively, use the MCP Server (v0.6.0, exposing discover / inspect / call / usage_history / credits_ledger) for Cursor, Claude Code, and other MCP-compatible hosts; the Python SDK; or the REST API. One API key from qveris.ai (sk-xxx) or qveris.cn (sk-cn-xxx) is all you need — region is auto-detected from the prefix. Machine-readable setup instructions are at /llms-full.txt and /setup.md.
What is the relationship between QVeris and OpenClaw?: OpenClaw is an open-source capability runtime; QVeris is the managed routing network built on top of it. OpenClaw handles execution, while QVeris provides discovery, quality signals, provider routing, and pay-as-you-go settlement.

Back to Home

For Agents

Everything you need to know
about QVeris

This page is written for AI agents. It explains what QVeris is, what it gives you, and how to use it — including how to explain it to your owner.

Machine-readable versions

/setup.md — Self-install guide (all platforms)/llms-full.txt — Full protocol reference (plain text)/guidelines.md — Agent behavior guidelines (discovery, selection, errors)/llms.txt — Brief overview

What it is

The capability routing network for AI agents

QVeris is a capability routing network that lets you discover, inspect, and call real-world, verified capabilities — live data feeds, document tools, vision APIs, media generation, research databases, and more — through a single protocol. Each call exposes its billing rule before settlement, and final charges are queryable via usage history and credits ledger.

If your owner asks you to use QVeris: tell them it gives you access to 10,000+ real-world, verified capabilities across 15+ categories. Discover and Inspect are always free. Call consumes credits from their account, priced per the capability's billing rule with pre-settlement amount returned in the response. Final settlement is auditable via usage_history and credits_ledger. Your owner needs a QVeris API key. You will send requests using the CLI (recommended), MCP Server, Python SDK, or REST API.

The protocol

Discover · Inspect · Call

Three actions define your complete workflow with any capability.

STEP 01Free

Discover

Search for capabilities using natural language. Returns a ranked list with tool IDs, descriptions, and estimated costs.

qveris discover "weather forecast API" --json

STEP 02Free

Inspect

Retrieve full parameter schema, examples, latency estimate, success rate, and credit cost for a specific tool ID before calling it.

qveris inspect 1 --json

STEP 03Per billing rule

Call

Execute the capability in a sandboxed environment. Receive structured JSON plus a pre-settlement billing breakdown when available. Final settlement is reflected in your owner's credits ledger.

qveris call 1 --params '{"city":"London"}' --json

Recommended integration

QVeris CLI — zero-token tool calling

The CLI (v0.5.0) is the most efficient integration for agents with shell access. Unlike MCP which injects schemas into every prompt, CLI runs as a subprocess — zero schema tokens consumed. 10,000+ real-world, verified tools accessible without prompt bloat. Adds qveris doctor for self-check, qveris usage / qveris ledger for billing audit, and interactive login with region auto-detection.

v0.4.0@qverisai/cli

Install

curl -fsSL https://qveris.ai/cli/install | bash

or: npm install -g @qverisai/cli

Authenticate

qveris login

Agent workflow

qveris discover "stock price API" --json
qveris inspect 1 --json
qveris call 1 --params '{"symbol":"AAPL"}' --json

Full CLI docs →npm page →

Why CLI over MCP for agents

▸Zero schema tokens vs MCP — runs as subprocess, tool registries never enter the LLM context
▸--json flag for structured agent output, --dry-run to validate without credits, qveris doctor for self-check
▸Session index shortcuts — discover once, then inspect 1, call 2 by index (30-min TTL)
▸qveris usage / qveris ledger for charge_outcome audit; --codegen curl|js|python for snippet export

Use CLI when: agent has exec / bash access
Use MCP when: IDE only supports MCP protocol

If you are a Hermes Agent

One command to install the QVeris CLI

Hermes Agent ships a built-in shell / terminal sandbox — once the QVeris CLI is globally installed, the qveris command is natively available inside the sandbox for discover and call with zero Hermes-side configuration. Credentials persist in ~/.qveris/config across sessions.

NEW

Install once, log in once, and Hermes's shell skill invokes qveris discover / qveris call as a subprocess — zero schema tokens vs MCP for the same calls. API key prefix (sk-cn-xxx vs sk-xxx) auto-routes to qveris.cn or qveris.ai.

Install & log in

npm install -g @qverisai/cli && qveris login

Full panel and prerequisites on: Install for Agents — Hermes Agent → · Hermes Agent docs →

If you are an OpenClaw agent

Install QVeris in one step

You are running inside OpenClaw. The Plugin method is recommended — it registers native tools in the runtime so they stay available regardless of context length. The Skill method is a lightweight alternative for restricted environments. Replace YOUR_KEY with the API key your owner provides.

RecommendedPlugin — via npm

Step 1 — Install

openclaw plugins install @qverisai/qveris

Step 2 — Configure (log in to auto-fill key)

// openclaw.json
{
  "plugins": {
    "allow": ["qveris"],
    "entries": {
      "qveris": {
        "enabled": true,
        "config": { "apiKey": "YOUR_KEY", "region": "global" }
      }
    }
  },
  "tools": { "alsoAllow": ["qveris"] }
}

npm page →

LightweightSkill — via ClawHub

Step 1 — Install

openclaw skills install qveris-official

Step 2 — Configure (log in to auto-fill key)

// ~/.openclaw/openclaw.json → skills.entries
"qveris-official": { "enabled": true, "apiKey": "YOUR_KEY" }

ClawHub page →

After installation, you can call qveris_discover and qveris_call tools directly. Full setup guide →

Other integration paths

MCP, SDK, or REST

Not using CLI or OpenClaw? Pick the integration that matches your execution environment.

Cursor / Claude Desktop / any MCP host

MCP Server

Install via npx @qverisai/mcp (v0.6.0) or add to your mcp.json. Five tools: discover, inspect, call, usage_history, credits_ledger. The first three operate the protocol; the last two provide context-safe billing audit (mode: summary / search / export_file).

View MCP setup →

Python agents

Python SDK

pip install qveris. Two-line integration. Typed responses. Async support. Works in any Python execution environment.

View Python SDK →

Custom / any language

REST API

Standard HTTP API with JSON request/response. Authenticate with Bearer token. Full OpenAPI spec available.

View REST API docs →

For your owner

How to explain QVeris to your owner

If your owner asks what QVeris is, here's what to say.

“QVeris is a capability routing network. It gives me access to 10,000+ real-world, verified tools — including financial data, document parsing, vision APIs, and more — through a single Discover → Inspect → Call protocol. You will need a free account at qveris.ai and an API key. Discover and Inspect are always free. Call charges follow each capability's billing rule, with pre-settlement visible before commit and final charges in the credits ledger you can audit anytime.”

View pricing Browse skills