Monetize your Code Execution MCP server metered compute is the original pay-per-use product.
Sandboxes, containers, GPU runtimes, even quantum backends — code execution servers sell compute by the call whether they charge for it or not. x402 makes the charging part real.
No category maps to pay-per-call more literally than code execution. Every tool call to a runno sandbox, a code-runner polyglot environment, or a podman-mcp-server container operation consumes a measurable slice of CPU, memory, and wall-clock time. Cloud providers have billed this exact resource by the unit for two decades; MCP maintainers mostly haven't.
The gap is infrastructure, not intent — and x402 closes it. A priced execution endpoint answers unpaid calls with HTTP 402, collects USDC before the sandbox spins up, and settles on Base in about two seconds. The agent pays before your handler runs, which in this category means before a single instruction executes.
Code Execution MCP servers on the Loomal Index
runno
MCP Server for the Runno Sandbox
code-runner
Code Runner MCP Server which can run code in various programming languages.
cesium-mcp-runtime
AI-powered 3D globe control via MCP. Camera, layers, markers, spatial analysis with CesiumJS.
podman-mcp-server
A Model Context Protocol (MCP) server for container runtimes (Podman and Docker)
Qiskit IBM Runtime MCP Server
MCP server for IBM Quantum computing services through Qiskit IBM Runtime
build123d MCP
AI-driven 3D CAD via build123d: execute, render, measure, and export geometry interactively.
glassbox-framework
Runtime constitutional verification for AI answers — claim reasoning, ECS, red team, audits.
RLM Tools
Persistent Python sandbox for token-efficient codebase exploration in MCP clients
mcp-server-colab-exec
Execute Python code on Google Colab GPU runtimes (T4/L4) from any MCP client
Vaara
EU AI Act runtime evidence proxy for MCP servers: gating, tamper-evident audit, time anchor
Vaara MCP Server
EU AI Act runtime evidence as a standalone MCP server: gating, tamper-evident audit
Shell Exec
Execute bash commands with background job support.
Showing 12 of 33 live Code Execution servers — browse them all on the marketplace.
Why execution must be paid to be offered at all
Free remote code execution is an abuse magnet — cryptominers and botnets find unmetered sandboxes within days. Most maintainers respond with API keys, quotas, and manual approval, which kills the spontaneous agent traffic they wanted. Payment is the cleaner filter: at even $0.01 per run, abusive workloads fund the infrastructure they consume or stay away.
The 33 live servers in this category are also the backbone of agent autonomy — an agent that can run code can verify its own work. That makes reliable, instantly-callable execution endpoints precisely what funded agents are out shopping for.
Pricing logic: runtime tiers, not flat rates
Price by what the execution occupies. Ephemeral sandboxed snippets — the runno model of safe, short-lived runs — work at $0.01–$0.05 per execution. Container lifecycle operations (build, run, manage via a podman-style server) sit similarly low per operation but add up across a workflow.
Specialized hardware is the premium tier and the clearest pricing story on the index: a GPU execution on a T4 or L4 through something like mcp-server-colab-exec has a knowable per-minute hardware cost — price the call above it and the margin is arithmetic. Exotic backends go higher still; quantum runtime access of the kind the Qiskit IBM Runtime server brokers is scarce enough that per-job prices in dollars are unremarkable. Long-running jobs deserve their own tools priced by duration class, so quick runs aren't subsidizing marathons.
Claiming and pricing your endpoint
Claim your listing on the Loomal index with a GitHub ownership check, then set a per-call price for each tool in the console — one field, adjustable whenever your hardware costs shift. Deploy the x402 middleware in front of your remote endpoint and you're gated: no payment, no execution.
Each settled call yields an Ed25519-signed receipt and a Base transaction hash. Settlement is final — no chargebacks after the compute is spent, which is exactly the failure mode card payments would expose you to. Loomal's fee is 5% on settled transactions, currently waived.
Safety and pricing are the same feature
In code execution, your isolation story is your product story. Buyers pay more for executions they trust: resource caps, network policies, clean teardown. Document those guarantees in your claimed listing — an agent (or its operator) choosing between execution endpoints will pay a premium for the one whose sandbox boundaries are explicit.
Per-call pricing reinforces the discipline. When every run is paid, you have both the means and the motive to keep isolation tight, capacity provisioned, and abuse priced out.
Frequently asked questions
Why charge for code execution instead of rate-limiting it?
Rate limits punish your best users and barely slow abusers with many identities. A per-call price — even Loomal's $0.01 minimum — makes every workload fund the infrastructure it consumes, while keeping the endpoint open to any agent with a wallet, no API-key provisioning required.
How do I price GPU or specialized-hardware executions?
Start from the hardware's known per-minute cost and price the call above it. A GPU run has a calculable floor; scarce backends like quantum runtimes can carry per-job prices in dollars. Expose duration classes as separate tools so short runs don't subsidize long ones.
What stops someone from running code and disputing the charge?
Nothing to dispute: x402 settlement in USDC on Base is final, with no chargeback mechanism, and payment completes before your handler executes. You also hold an Ed25519-signed receipt and an on-chain transaction hash for every call.
How do I get my execution server earning?
Claim its listing on Loomal, verify ownership via GitHub, set per-call prices in the console, and put the x402 middleware in front of your remote endpoint. Loomal charges 5% on settled transactions, a fee that's currently waived.
Run a Code Execution MCP server?
Claim your listing, set a per-call USDC price, and let AI agents pay for every call over x402.
List it on Loomal