QVeris Agent Toolkit MCP
The MCP server, CLI and SDKs of the QVeris platform: the agent finds a data service, calls it and audits credit spending
- 263 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- `call` invocations are billed and spend your account credits, so check the price before running
- Requests and their parameters go to a remote platform and third-party data providers
- The `QVERIS_API_KEY` sits in the client environment during a local run
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/qveris-agent-toolkitDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- `call` invocations are billed and spend your account credits, so check the price before running
- Requests and their parameters go to a remote platform and third-party data providers
- The `QVERIS_API_KEY` sits in the client environment during a local run
This is third-party code. Review the repository files before installing.
What it does
QVeris is a platform of professional data and tools for AI, and this repository gathers its open clients: the `@qverisai/mcp` MCP server, the `@qverisai/cli` CLI, Python and TypeScript SDKs, skills and REST API docs. The agent does not receive a catalog of hundreds of services upfront. It finds the right one from a natural language request, checks the terms and calls it. The work cycle has several steps. `discover` searches for suitable services and is free. `inspect` shows missing parameters, supported scope and billing when needed. `probe` validates parameters and a quote without executing. `call` runs the chosen service and returns a structured result with a pre-settlement cost estimate. The other tools, `usage_history`, `credits_ledger` and `lookup_execution`, serve auditing: final charges, credit movement and result lookup by idempotency key. There are two ways to connect. The hosted option takes the address `https://mcp.qveris.ai/mcp`, sign-in goes through the browser by OAuth and nothing needs installing. A local run goes through `npx -y @qverisai/mcp` with the `QVERIS_API_KEY` variable. Cursor has a marketplace plugin, Gemini CLI an extension and OpenClaw an official skill. The platform covers finance, search, weather, maps, documents, social, blockchain and healthcare. The catalog of recipes in the `recipes/` folder shows ready scenarios for financial analysis, crypto monitoring and developer automation.
Who it is for. For agent developers and people working in Claude Code, Cursor or Codex who want to connect professional data sources on demand without configuring each service separately.
Good fit when
- The agent's built-in tools fall short, for example for market data, weather or maps, and you want one search across services
- You need to compare providers for one task: `inspect` and `probe` show scope and a current quote for each candidate
- Spending audit matters: `usage_history` and `credits_ledger` give a history of calls and credit charges
- You are writing your own app: there are Python and TypeScript SDKs and a REST API with the same protocol steps
- You need a shell run without loading the whole catalog into context: `qveris discover`, `inspect` and `call` in a terminal
Not a fit when
- Local or built-in tools already solve the task: the documentation itself advises using them first
- You need a fully free and standalone source: `call` invocations are billed, only `discover` is free
- Company policy forbids sending requests to an external platform or forbids cloud data services
- You need a fixed set of services with guaranteed terms: the platform picks services dynamically
Example request
Through QVeris find a service with the current EUR to USD rate, show the terms and cost through `probe`, and only then call it.MCP
- Transport
- http, stdio
- Authentication
- OAuth
Environment variables
QVERIS_API_KEYsecret- API key, needed only for a local stdio run or clients without OAuth support
QVERIS_BASE_URL- Optional base API address, https://qveris.ai/api/v1 by default
Limitations
- A QVeris account is required. The trial package is 1000 one-time credits after signup verification, then pay-as-you-go applies.
- A call's price is set by the service's structured billing rule. A pre-settlement estimate arrives in the `call` response, and the final charge is visible through `usage_history` and `credits_ledger`.
- Requests and parameters go to a remote platform, while called services get no access to your file system and settings.
- A local run through `npx` needs Node.js 18 or newer and a `QVERIS_API_KEY`. The hosted MCP with OAuth needs no key.
- The old tool names `search_tools`, `get_tools_by_ids` and `execute_tool` are supported as deprecated aliases.
How to disable. Delete the `qveris` entry from the MCP client config. For Gemini CLI remove the extension, for Cursor disable the plugin. You can revoke the API key on the QVeris account page and remove the CLI with `npm uninstall -g @qverisai/cli`.
FAQ
How do I connect QVeris to Claude Code?
Add an HTTP server with the address `https://mcp.qveris.ai/mcp` to the MCP config and sign in through the browser. If the client cannot do OAuth, run `npx -y @qverisai/mcp` with the `QVERIS_API_KEY` key.
How much do QVeris calls cost?
The `discover` search is free, while `call` is billed under the service's rule. 1000 trial credits are given after signup verification. `inspect`, `probe` and the pre-settlement estimate in the response show the exact price.
How does inspect differ from probe?
`inspect` shows a service's parameters, scope and billing terms. `probe` validates specific parameters and gives a current quote without running the call.
How do I verify what was charged for a call?
Use `usage_history` for the request history and `credits_ledger` for the final credit movement. In the CLI the `qveris usage` and `qveris ledger` commands return the same data.
What if a call hangs and the result is unclear?
`lookup_execution` finds the outcome by the idempotency key you passed to `call`. It only reads and repeats nothing.
Discussion
No questions or reviews yet.
Related
MCP serversAn MCP server with a toolbox for global briefs: markets, SEC filings, conflict, climate and news from open sources
A SQLite-based SQL engine with plugins for apps, databases and files: the anyquery mcp command exposes that data to Claude, Cursor and other MCP clients
MCP server and CLI to drive real Microsoft Excel: Power Query, DAX, VBA, PivotTables and charts
An MCP server from AntV for generating charts and diagrams: 25+ visualization types, from lines and bars to maps, mind maps and network graphs
Unofficial Checko.ru MCP server for counterparty checks: EGRUL, EGRIP, financials, court cases, state contracts and bankruptcies
A local MCP server that turns a folder of documents into a knowledge base with hybrid search for Claude Code and Cursor
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| QVeris Agent Toolkit (this entry) | yes | OAuth | 263 | |
| Jarvis Registry | yes | OAuth | 3,702 | a self-hosted MCP and A2A gateway for your own infrastructure, while QVeris is a hosted platform with a catalog of data services |
| Membrane Application Skills | no | 270 | teaches an agent to work with Gmail, Slack, HubSpot and Salesforce, while QVeris searches professional data sources |
README in short
The page describes the platform and gives a quick start for agents, OpenClaw, Cursor and Claude Code users and developers. It then covers rules for `discover`, `call`, `inspect` and `probe`, billing, the CLI and package tables.