Integration Guide

Integration Surfaces

ATF provides four integration surfaces: a hosted MCP endpoint for agent runtimes, a REST API for direct integration, a CLI for local development and CI pipelines, and an OpenClaw plugin for autonomous agent frameworks.

Surface Overview

Hosted MCP

Available

Model Context Protocol endpoint for agent runtimes. Five tools covering advisory probing, simulation, protect enforcement, verification, and explanation.

REST API

Available

The primary integration path. Send intents, receive decisions, verify receipts. Works with any language or framework.

CLI

Available

Local development, scripting, and CI/CD pipelines. Profile-based config, pipeline-friendly I/O.

OpenClaw Plugin

Available

Autonomous agent integration. 13 tools for protect, verify, report, and discover. Runs inside the OpenClaw agent framework.

Hosted MCP Endpoint

ATF exposes a hosted Model Context Protocol (MCP) endpoint for agent runtimes. Five tools cover the full advisory-to-enforcement loop. Advisory tools (probe_transaction, simulate_transaction, explain_decision) are policy-aware but not authoritative. protect_transaction is the binding enforcement gate. MCP does not sign or submit transactions.

Entitlement is tier-based and tenant-backed. Each tenant's MCP access is scoped to their entitlement tier.

Tool	Status	Description
probe_transaction	Available	Lightweight policy pre-check on a candidate intent (advisory)
simulate_transaction	Available	Full simulation against active policies and conditions (advisory)
protect_transaction	Available	Binding policy enforcement decision - approve or deny (authoritative)
verify_receipt	Available	Verify execution receipt hash integrity
explain_decision	Available	Human-readable explanation of a decision with reason codes (advisory)

MCP Boundaries

MCP does not sign transactions
MCP does not submit transactions to the chain
Advisory tools inform; protect enforces
Entitlement is tier-based and tenant-backed

Canonical MCP flow

Discover tools via MCP endpoint
Probe a candidate intent
Simulate the candidate
Request protection (binding enforcement)
Verify the execution receipt
Explain the result if needed
Stop before signing/submission

REST API

Base URL: https://api.trucore.xyz. All endpoints accept JSON. Authentication is via X-API-Key header for protect/execute operations.

bash

curl -sS https://api.trucore.xyz/v1/bot/protect \
  -H "Content-Type: application/json" \
  -H "X-API-Key: atf_live_YOUR_KEY" \
  -d '{
    "chain_id": "solana",
    "intent_type": "swap",
    "intent": {
      "type": "swap",
      "in_mint": "So11111111111111111111111111111111111111112",
      "out_mint": "EPjFWdd5AufqSSqeM2qN1xzybapC8G4wEGGkZwyTDt1v",
      "amount_in": 1000000,
      "slippage_bps": 50,
      "agent_id": "my-bot-v1"
    }
  }'

Endpoint	Status	Description
POST /v1/bot/protect	Available	Evaluate intent against policies, return allow/deny with receipt
POST /v1/intents/approve	Available	Submit swap, lending, or withdrawal intent for approval
POST /v1/executions/finalize	Available	Finalize an approved execution and issue a full receipt
POST /v1/receipts/verify	Available	Verify receipt hash integrity
POST /v1/permits/verify	Available	Verify permit signature and expiration
GET /v1/whoami	Available	Resolved identity and tenant metadata
GET /plans	Available	Public plan tiers and limits
GET /features/catalog	Available	Public-safe feature catalog
GET /onboarding/sample-intent	Available	Pre-built sample swap intent for testing
POST /onboarding/protect-dry-run	Available	Policy evaluation without on-chain execution
POST /onboarding/execute-sample	Available	Execute sample trade (mock or real)

See the full API Reference for request/response schemas and error codes.

ATF CLI

Install via npm and run immediately. The CLI wraps the API with profile management, local receipt verification, and pipeline-friendly exit codes.

bash

# Install globally
npm install -g @trucore/atf

# Run a protected trade (demo mode)
atf trade

# Protect a specific intent
cat intent.json | atf bot protect --stdin

# Verify a receipt
atf verify <receipt-id>

# Environment health check
atf doctor

Command	Status	Description
atf trade	Available	Run a protected trade (demo or real)
atf setup	Available	Configure wallet and RPC for real trades
atf doctor	Available	Environment health check - RPC, wallet, config
atf verify	Available	Verify receipt hash locally
atf simulate	Available	Simulate a transaction without sending
atf bot protect	Available	Protect intent from stdin (pipeline-friendly)
atf receipts list	Available	List receipts for current profile
atf profile	Available	Manage named profiles for different environments
atf rpc ping	Available	Test RPC endpoint latency
atf burner	Available	Switch active profile to devnet for testing
atf whoami	Available	Show current identity and profile

See the full CLI Reference and Command Reference for details.

OpenClaw Plugin

The @trucore/trucore-atf plugin integrates ATF into the OpenClaw autonomous agent framework. Agents can protect intents, verify receipts, and generate savings reports without custom HTTP code.

bash

openclaw plugins install @trucore/trucore-atf@0.2.11

Tool	Status	Description
atf_health	Available	Check ATF service health
atf_discover	Available	Discover ATF capabilities and manifest
atf_protect_intent	Available	Submit intent for policy evaluation
atf_verify_receipt	Available	Verify receipt integrity
atf_report_savings	Available	Generate receipts-backed savings report
atf_bot_preflight	Available	Run bot preflight checks
atf_integration_doctor	Available	Diagnose integration issues
atf_tx_explain	Available	Explain a transaction or intent
atf_adoption_advisor	Available	Adoption guidance and recommendations
atf_billing_status	Coming Soon	Check billing and usage status
atf_billing_upgrade	Coming Soon	Trigger plan upgrade flow

See the full OpenClaw Plugin Reference for configuration, safety defaults, and workflow examples.

When to Use Which Surface

Use MCP when…

Your agent runtime supports MCP natively
You want the advisory-to-enforcement loop in one surface
Building with hosted agent frameworks

Use the API when…

Building a custom trading bot in any language
Integrating ATF into an existing backend
You need full control over request/response handling

Use the CLI when…

Developing and testing locally
Running in CI/CD pipelines
Scripting protect/verify workflows

Use the Plugin when…

Building on the OpenClaw agent framework
Want zero-code ATF integration for agents
Need autonomous protect → verify workflows