DLL Execution MCP server
Commit Oct 22, 2026Allows AI agents to simulate Base Mainnet transactions and verify on-chain execution outcomes without requiring signing or custody.
Also known as dllio/dll-execution-docs, dll-execution-docs
- Stars
- 0
- Tools
- —
- Downloads, 30 days
- —
- Latest version
- 0.1.0
- Last commit
- Oct 22, 2026
- License
- MIT
What is DLL Execution MCP?
DLL Execution allows AI agents to simulate Base Mainnet transactions and verify on-chain execution outcomes without requiring signing or custody.
Summary generated from the project's README
About DLL Execution MCP server
DLL Execution provides a read-only infrastructure for AI agents to interact with the Base Mainnet without requiring wallet connections, signatures, or custody. It functions as a remote MCP service or a direct API, allowing developers to integrate transaction simulation and post-execution observation into their agent workflows. The service is designed to help agents understand potential failures, estimated costs, and execution constraints before any funds are moved.
The system offers three primary capabilities: Preflight, Verification, and Benchmark. Preflight evaluates unsigned transaction candidates to provide structured judgments like PASS, CAUTION, or BLOCK. Verification observes already-broadcast transaction hashes to confirm inclusion and receipt status. Benchmark compares transaction costs against historical data using a specific payment model. Because the service is read-only, it never signs or broadcasts transactions, ensuring that the agent maintains full control over its private keys and assets.
What you can do with it
- 1Simulate unsigned transactionsAn AI agent can evaluate a transaction candidate to check for potential reverts or fee constraints before signing.
- 2Verify transaction outcomesA developer can instruct an agent to check if a previously broadcast transaction was successfully included on the Base Mainnet.
- 3Analyze historical transaction costsAn agent can use the benchmark tool to compare the costs of current transactions against historical context.
Written from the project's README. Every use case is checked against a quote from it.
How to install DLL Execution MCP
Formats verified 10 Nov 2026Add to CursorAdd to GooseInstall DLL Execution MCP in Claude Code
$ claude mcp add --transport http dll-execution-docs https://execution.dll.io/mcpInstall DLL Execution MCP in Cursor
~/.cursor/mcp.json
{
"mcpServers": {
"dll-execution-docs": {
"url": "https://execution.dll.io/mcp"
}
}
}Install DLL Execution MCP in VS Code
.vscode/mcp.json
{
"servers": {
"dll-execution-docs": {
"type": "http",
"url": "https://execution.dll.io/mcp"
}
}
}Install DLL Execution MCP in VS Code CLI
$ code --add-mcp '{"name":"dll-execution-docs","type":"http","url":"https://execution.dll.io/mcp"}'Install DLL Execution MCP in Codex
$ codex mcp add dll-execution-docs --url https://execution.dll.io/mcpInstall DLL Execution MCP in Codex config.toml
~/.codex/config.toml
[mcp_servers.dll-execution-docs] url = "https://execution.dll.io/mcp"
Install DLL Execution MCP in Gemini CLI
$ gemini mcp add -s user --transport http dll-execution-docs https://execution.dll.io/mcpInstall DLL Execution MCP in Gemini settings.json
~/.gemini/settings.json
{
"mcpServers": {
"dll-execution-docs": {
"url": "https://execution.dll.io/mcp",
"type": "http"
}
}
}Automated checks
Ran Nov 15, 2026Full report →Not verified yet. A server needs a score of 80, no serious finding, and enough data for us to check. How checks work · What to fix for more points
- Source repository linked and reachabledllio/dll-execution-docs is public and reachablegithub.com/dllio/dll-execution-docs ↗Pass
- Package names this repository as its sourceNot checked
- Build provenance (Sigstore attestation)Not checked
- Open source license (OSI)MITspdx.org/licenses/MIT.html ↗Pass
- Commit or release in the last yearLast commit or release 2026-10-22github.com/dllio/dll-execution-docs ↗Pass
- OpenSSF ScorecardNot checked
- Package name not close to a popular packageNot checked
- No critical advisory in the package or its dependenciesNot checked
- No hidden instructions to the model0 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
- Destructive tools are labelledNot checked
- Remote endpoint answers the MCP handshakeEndpoint did not answer initialize: timed outFinding
README
View on GitHubDLL Execution — Base transaction simulation and verification for AI agents
Execution infrastructure for AI agents.
DLL is an independent, read-only Base Mainnet API and remote MCP service. Before signing, simulate an unsigned transaction, inspect call success or revert, and check explicit destination and L2 execution fee constraints. After independent broadcast, verify inclusion, requested receipt status and supported transaction-field expectations. Results include structured Judgment or Verification states, reason codes, evidence and explicit uncertainty; DLL never signs, broadcasts or takes custody.
Who it is for
AI Agents with an unsigned transaction candidate or an already-broadcast transaction hash, and developers integrating those Agents. Before acting, an Agent needs to understand likely failure, known costs, explicit constraints and remaining uncertainty. Afterward, it needs to check observed outcomes against explicit expectations.
Available now: Execution Preflight
POST https://execution.dll.io/v2/execution/preflight
V0.2.1 evaluates one unsigned Base Mainnet transaction candidate and returns read-only simulation, an estimated L2 execution fee, deterministic policy checks, structured judgment and supporting evidence. It is a limited free public trial; no wallet connection, signature or payment is needed for Preflight. Benchmark's separate payment/free-trial rules do not apply to this endpoint.
With complete supported EIP-1559 transaction context, optional fee_estimation returns a separate conditional estimate of L1 data, L2 execution, applicable Operator Fee and Total Network Fee. Missing necessary components keep total unavailable, not zero. The estimate is not realized cost, a future spending guarantee or an automatic PASS.
Agent has an unsigned transaction candidate
→ calls DLL Execution Preflight
→ receives simulation / cost / policy / judgment / evidence
→ Agent decides what to do next
| Judgment | Meaning within the evaluated evidence |
|---|---|
PASS |
Simulation and applicable checks passed, with no material unresolved issue identified. Not permission to execute. |
CAUTION |
Execution may be technically possible, but important uncertainty or incomplete evidence remains. |
BLOCK |
Deterministic evidence, such as a simulation revert or explicit constraint rejection, indicates not to proceed with this candidate. |
INDETERMINATE |
Available evidence cannot support a reliable judgment. Do not treat it as success or a proven transaction failure. |
Ordinary successful Preflight currently returns CAUTION because legacy cost_estimate L1/total remain unavailable. A complete optional network_fee_estimate does not recompute legacy Judgment or the L2-only cost constraint. None of these states evaluates profitability, investment quality, strategic desirability or financial advice.
Available now: Execution Verification
POST https://execution.dll.io/v2/execution/verify
Verification V0.1 observes an already-broadcast Base Mainnet transaction and evaluates narrow expectations about its chain, inclusion, receipt status and optional candidate content. It is free and read-only: no signature, wallet connection or payment is required, and DLL need not have performed the original Preflight.
- Preflight — before execution: evaluate what is likely to happen, within the recorded evidence.
- Verification — after execution: observe what actually happened and evaluate it against explicit expectations.
Only transaction_hash is required. The default checks Base inclusion; request receipt_status: "success" explicitly to check receipt success. Read verification.state, not just HTTP status: VERIFIED means the requested checks are satisfied, CONTRADICTED means at least one check is contradicted, and INDETERMINATE means evidence is insufficient. These states do not establish finality, authorization, ownership, causality or profitability.
Available now: Execution-cost Benchmark
POST https://execution.dll.io/v1/benchmark/transactions
Benchmark compares observed Base Mainnet transaction costs with historical context. The public path uses SIWX wallet identity and x402 v2 upto USDC: the first three valid analyses per verified wallet are free, then $0.005 per delivered valid analysis. Anonymous nonempty payment quotes do not force free-eligible wallets to pay. A paid N-hash batch quotes a maximum of N × 5000 atomic USDC; actual settlement counts only valid paid items. HTTP 200 can contain unavailable items; inspect completed billing. Historical percentiles are not route recommendations, profitability judgments or future fee guarantees. See the Benchmark guide and bounded SDK client reference.
Start here
Start with HTTP or Remote MCP Quickstart: no DLL account, API key, wallet or private repository access. The JavaScript path uses only Node.js built-ins; the remote MCP path uses your existing MCP client. No npm install or AgentKit SDK is required.
- Remote MCP: two free read-only tools and standard client setup; no wallet or payment.
- Capability selection and FAQ: choose Preflight, Verification or Benchmark and understand their boundaries.
- Quickstart: make your first Preflight or Verification call.
- Agent integration instructions: when to call and how to interpret results.
- Preflight contract: inputs, outputs, constraints and errors.
- Verification contract: inclusion, receipt/content checks, realized costs, observation states and errors.
- Benchmark contract: historical cost context, SIWX, allowance, x402 payment and safe reconciliation.
- Benchmark client reference: stage-aware identity/payment negotiation with independent spending authorization, not a generic-client compatibility guarantee.
- Preflight examples: curl, JavaScript, Python.
- Verification examples: curl, JavaScript, Python.
Public discovery and canonical contract
- API origin: compact capability selector and canonical integration links.
- Machine-readable index: index of authoritative public surfaces, not a duplicate API contract.
- Public sitemap and crawl hints: explicitly advertised public GET documents; not authentication or proof of indexing.
- Live OpenAPI: canonical API schemas and available operations.
- Capabilities: currently available capabilities and their scope.
- x402 discovery manifest: Benchmark paid resource discovery; neither Preflight nor Verification requires x402 payment.
This repository explains what DLL does and how to use it. Live OpenAPI, well-known metadata and actual API responses are the canonical public contract. It is not an implementation/source-code mirror.
Documentation checks run offline with node --test test/*.test.mjs: local links, JSON syntax and bounded client safety behavior. They create no real signatures, payments or blockchain requests. Full examples are additionally checked against canonical product schemas during release validation.
Limitations and security boundary
Base Mainnet only. Preflight accepts unsigned candidates; Verification accepts already-broadcast transaction hashes. No signing, broadcasting, custody or autonomous execution. DLL does not prove ownership of the supplied from address or authorize a transaction.
Simulation describes observed chain state; it does not guarantee future inclusion, success or application-level outcomes. The default L2-only fee is not total execution cost; optional network fee completeness describes only the supported conditional model. An unavailable fee is unknown, not zero. DLL does not verify net asset changes, comprehensively scan malicious contracts or guarantee Swap economic results. Keep private keys, credentials and other secrets out of requests. The Agent remains responsible for any later action.
Verification checks only the supplied expectations. Included, successful and canonical observations are not finalized-state guarantees. Its realized L2 fee is not total cost; total execution cost currently remains unavailable even if an L1 fee is observed.
DLL Execution MCP questions
How do I add DLL Execution MCP to Claude Code?
Run claude mcp add --transport http dll-execution-docs https://execution.dll.io/mcp in your terminal. Add --scope user to make it available in every project.
How do I add DLL Execution MCP to Cursor?
Open ~/.cursor/mcp.json and add the server under mcpServers, using the Cursor config shown in the install section above.
Does DLL Execution handle private keys or signing?
No, the service is read-only and never signs, broadcasts, or takes custody of assets.
What is the difference between Preflight and Verification?
Preflight evaluates an unsigned transaction candidate before execution to predict success and costs, while Verification observes an already-broadcast transaction to check its actual outcome.
Is there a cost to use the Preflight or Verification tools?
Preflight and Verification are free, read-only services that do not require a wallet connection or payment.