Open Source AI Agents
On this page
Open source AI agents are agent frameworks whose code you can read, run, and modify, and they reach external APIs through tool interfaces. The Model Context Protocol (MCP) is the common standard for those tool connections, so an MCP gateway lets an open source agent call many APIs through one endpoint. Pick the framework for control of the agent loop, then pick the tool layer for breadth of APIs.
What makes an AI agent open source
An open source AI agent publishes its source code under a license that lets you inspect, run, and change it. The agent loop, the prompt handling, and the tool-calling logic all live in files you can read. A closed agent hides that logic behind a hosted product.
Open source does not mean free to run. The agent still calls a large language model (LLM), and that model bills per token or runs on your own hardware. The framework cost is zero, but the model and the tools cost money or compute.
How open source agents call APIs
An agent reaches an external API through a tool. The tool wraps one HTTP request, sends JSON parameters, and returns a JSON response that the model reads.
MCP standardises that wrapper. An MCP server exposes a list of tools with typed inputs, and any MCP-aware client calls them the same way. The agent does not need custom glue code for each API.
- The client discovers the tool list from the server.
- The model picks a tool and fills its JSON arguments.
- The server runs the HTTP call and returns the result.
Choose a framework and a tool layer
Choose the framework for how the agent plans and loops. Choose the tool layer for which APIs the agent can reach. The two choices are independent. You can swap one without rebuilding the other.
Frameworks differ in language, state handling, and multi-agent support. Tool layers differ in how many APIs they expose, how they handle authentication, and how they bill calls. A gateway with a large and growing catalog saves you from writing one connector per API.
Pricing and payment for agent tool calls
Each tool call costs money when the upstream API charges for it. The agent pays per call, not per seat. Track the cost per task, because a loop that retries a paid tool multiplies the bill.
APIbase tool calls are priced per call. Individual tools range from 0 to 29.99 per call. Check the current price on each tool page before you set a budget.
Security and control for open source agents
An open source agent runs code you control, so you own the secrets. Store API keys in environment variables or a secrets manager, never in prompts or in the repository.
- Give each agent a scoped key with only the tools it needs.
- Cap spend per run and log every tool call.
- Review the tool descriptions, because the model chooses tools from them.
Getting started with an open source agent
Start with one framework, one model, and two or three tools. Run a single task end to end before you add more tools.
Connect the agent to an MCP gateway with one endpoint. Then add tools as the task requires. Keep the agent loop small until the first task completes reliably.
Live pricing — developer
Data source: live tool catalog snapshot, fetched 2026-11-19.
Chart data as a table
| Tool | Price per call | Cache-hit price |
|---|---|---|
| Usage Time Series | $0.0 | $0.0 |
| Per-Tool Usage Breakdown | $0.0 | $0.0 |
| Usage Summary | $0.0 | $0.0 |
| Discover Tools | $0.0 | $0.0 |
| Batch Tool Calls | $0.0 | $0.0 |
| Tool Quality Metrics | $0.0 | $0.0 |
| Tool Quality Rankings | $0.0 | $0.0 |
| List Programming Languages | $0.001 | $0.0001 |
| Check CVE ID Reservation Status | $0.001 | $0.0001 |
| Security Advisories (deps.dev) | $0.001 | $0.0001 |
| Dependency Tree (deps.dev) | $0.001 | $0.0001 |
| Package Info (deps.dev) | $0.001 | $0.0001 |
| Tool | Provider | Price/call | Cache-hit |
|---|---|---|---|
| Usage Time Series | account | $0 | $0 |
| Per-Tool Usage Breakdown | account | $0 | $0 |
| Usage Summary | account | $0 | $0 |
| Discover Tools | apibase | $0 | $0 |
| Batch Tool Calls | platform | $0 | $0 |
| Tool Quality Metrics | platform | $0 | $0 |
| Tool Quality Rankings | platform | $0 | $0 |
| List Programming Languages | judge0 | $0.001 | $0.0001 |
| Check CVE ID Reservation Status | cve-mitre | $0.001 | $0.0001 |
| Security Advisories (deps.dev) | depsdev | $0.001 | $0.0001 |
| Dependency Tree (deps.dev) | depsdev | $0.001 | $0.0001 |
| Package Info (deps.dev) | depsdev | $0.001 | $0.0001 |
Connect via MCP
Data source: gateway design, no measured values.
$ curl -X POST https://apibase.pro/api/v1/tools/account.timeseries/call \
-H "Content-Type: application/json" -d '{"params": {}}'FAQ
Are open source AI agents free to use?
The framework code costs nothing to use. The language model and the paid API tools still bill per use. Budget for model tokens and tool calls separately.
What is MCP in an open source agent?
MCP is the Model Context Protocol. It defines how an agent discovers and calls tools on an MCP server, so one client can use many APIs without custom connectors.
Can an open source agent call paid APIs?
Yes. The agent sends the request with the key you provide. The upstream provider charges for each call, so set spend limits before you run loops.
How do I pick an open source agent framework?
Pick by language, state handling, and multi-agent needs. Test two frameworks on the same task. Keep the one that completes it with fewer retries.
Recommended next step
- STEP 1/2 Connect your MCP client
- STEP 2/2 Browse the full catalog
Related guides
Written with AI assistance from APIbase's published documentation and public data; checked automatically against them. Corrections: https://apibase.pro/
Published: 2026-11-19Source: live apibase.pro catalog snapshot