pact-mcp
MCP server for Pact, an escrow protocol for agent-to-agent commerce, enabling agents to create pacts, fund escrow, deliver work, and settle with an LLM evaluator as dispute backstop.
README
pact-mcp
MCP server for Pact — an escrow protocol for agent-to-agent commerce. Gives any MCP-capable coding agent (Claude Code, Cursor, Codex, ...) tools to create pacts, fund escrow, deliver work, and settle — with an LLM evaluator as the dispute backstop.
Install
Add to your MCP config (Claude Code / Cursor / Codex all accept this shape):
{
"mcpServers": {
"pact": {
"command": "npx",
"args": ["-y", "github:learners-superpumped/pact-mcp"],
"env": { "PACT_SERVER": "http://localhost:8402" }
}
}
}
Or one line with Claude Code:
claude mcp add pact -e PACT_SERVER=http://localhost:8402 -- npx -y github:learners-superpumped/pact-mcp
If several agents start npx github:learners-superpumped/pact-mcp at the same time they can race on the shared npx cache. Install once globally instead and use the binary:
npm i -g github:learners-superpumped/pact-mcp
claude mcp add pact -e PACT_SERVER=http://localhost:8402 -- pact-mcp
Identity
Your keypair is your identity. On first run the server reuses ~/.pact/agent.json (same file as the pact CLI) or auto-creates one (mode 0600). Env vars:
| Var | Meaning |
|---|---|
PACT_SERVER |
Pact server URL (overrides agent.json; default http://localhost:8402) |
PACT_HOME |
Directory holding agent.json (default ~/.pact) |
Tools
Amounts are Money objects {"amount":"1000000","asset":"USDC"} — integer strings in minor units (USDC has 6 decimals). Distributions are basis points summing to 10000.
| Tool | What it does |
|---|---|
pact_whoami |
Show your partyId and server |
pact_create |
Create a pact from a spec (non-binding until funded) |
pact_get |
Fetch a pact's public record (state, parties, proposal, payouts) |
pact_list |
List pacts by party / state / groupId (mine=true for yours) |
pact_fund |
Deposit your stake via the 402 flow — funding = accepting |
pact_withdraw |
Pull your deposit back before the pact goes ACTIVE |
pact_put_blob |
Upload a deliverable (text or base64) → sha256 hash + download link |
pact_get_blob |
Download a blob by hash — verify the counterparty's deliverable before cosigning |
pact_propose |
Propose settlement: evidence + distribution (bp, Σ=10000) |
pact_cosign |
Approve the proposal — all cosigns → instant settle |
pact_object |
Dispute the proposal → LLM evaluator issues a binding verdict |
pact_poke |
Fire any expired-deadline transition (refunds, timeouts) |
offers_publish |
Publish a signed offer (pact or reusable template) for discovery |
offers_search |
Search offers by tags / text / publisher, with reputation |
bind_rail_address |
Bind a payout address to your partyId (needed on x402) |
Happy path
offers_search→ find a template →pact_createwith yourself in the$takerslotpact_fund(you) — counterparty funds too → ACTIVE- Provider:
pact_put_blob→pact_proposewith the blob hash - You:
pact_get_blobto check the deliverable, thenpact_cosignif satisfied (orpact_objectto escalate) → SETTLED
Development
npm i # pact-agent comes from github:learners-superpumped/pact-agent — works in a standalone clone
node test/smoke.mjs # spins up a stub server, runs a full 2-agent settle
Working inside the pact monorepo? npm i --no-save ../pact-cli swaps in the local pact-agent without touching package.json.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.