Pay-per-call x402 gateway for AI agents — MCP server + OpenAI-compatible API for agent tools, market data, RPC, and MCP security audits. Pay in USDC on Base, no account.
mcp-name: com.daedalusdevelopmentgroup/ddg-agent-services-mcp
以下摘录自该仓库的 README(片段) · 在 GitHub 读完整 README →
DDG Agent-Payable Revenue Launch Pack
This directory contains the machine-readable launch artifacts and local scaffold for turning Hermes/DDG into a paid service surface for other AI agents.
Current launch wedges
1. **One-cent transaction smoke test** (/v1/tx-smoke-test) — $0.01 receipt-only paid request so other AI agents can prove they can complete a DDG paid HTTP transaction.
2. **Website audit** (/v1/site-audit) — read-only; launch settlement-test price $0.75.
3. **Model router / basic conversation** (/v1/model/chat-completions, /v1/model/agent-run) — paid access to DDG-operated model routing using explicit model aliases; supports basic chat, analysis, code/help prompts, and bounded agent tasks.
4. **Operator-reviewed OAuth/model artifact run** (service_id=provider_model_artifact_order through /v1/order-intake) — DDG-operated account/provider-backed model work only where terms permit; no raw token/account/seat resale.
5. **Lead pack** (/v1/lead-pack) — paid local business intelligence; launch price $5.
6. **Browser proof** (/v1/browser-proof) — paid screenshots/QA proof; launch price $2.
7. **Outreach brief** (/v1/outreach-brief) — paid draft only; no sending without human approval; launch price $1.50.
8. **Ethereum private RPC query** (/v1/ethereum/rpc) — implemented read-only, method-whitelisted Reth proxy with sync gate, fair-use lease timer, per-agent request caps, max batch size, and upstream max-concurrency cap; public self-serve waits for execution sync/health.
9. **Local model free-seat pool** (/v1/micro-model-swarm-preview, /v1/model/agent-run, /v1/ollama-models) — live free/best-effort Ollama local routes with one concurrent generation seat per healthy configured local runtime backend, capped at two by default for configured 8GB-VRAM backends. Defaults: 24-hour per-agent lease, 100 requests/lease, 60s swarm generation timeout, 120s local agent-run timeout. The named paid lease path remains planned.
10. **Agent artifact/receipt console** (/v1/orders/{order_id}, /v1/orders/{order_id}/artifact) — live agent-scoped status/artifact URLs returned from /v1/order-intake; exposes redacted receipt/proof hashes and expected deliverables, never raw payment material.
11. **Checkout conformance profile/probe** (/.well-known/ddg-agent-checkout-conformance.json, scripts/agent_checkout_conformance_probe.py) — lets buyer agents or partner services validate identity gating, 402 challenges, fail-closed fake payments, idempotency, accepted order links, and artifact access controls.
12. **MPP activation guide** (/.well-known/ddg-mpp-activation-guide.md) — operator/buyer-facing account creation, Stripe/Tempo/mppx test setup, required DDG secrets, live gates, public-copy flip checklist, and rollback steps.
Penny transaction smoke test
POST /v1/tx-smoke-test is the tiny paid-request endpoint. It is meant for other AI agents' CI/smoke tests: get a normal 402 challenge, pay one cent through x402 or direct-crypto now; MPP appears only after live provider/currency config, retry, and receive a receipt-only JSON response. It does **not** perform any irreversible external action beyond the payment verification/settlement itself.
Ethereum/Reth privacy-node posture
The /v1/ethereum/rpc route is for accountless buyer-side Ethereum reads: DDG runs the Reth node/proxy, the buyer agent does not need an Infura/Alchemy/vendor account, and the edge still requires AI-agent identity. The proxy is read-only/method-whitelisted; admin, miner, personal, unrestricted debug/trace, and eth_sendRawTransaction are blocked by default. The free beta wrapper now includes:
- sync readiness gate: DDG_ETH_RPC_REQUIRE_SYNCED=1 returns 503 reth_rpc_syncing until eth_syncing=false and latest block is non-zero/current;
- fair-use timer: DDG_ETH_RPC_FREE_MAX_CONCURRENT_USERS=4, DDG_ETH_RPC_FREE_LEASE_SECONDS=600, DDG_ETH_RPC_FREE_MAX_REQUESTS_PER_LEASE=30 by default;
- node overload guard: DDG_ETH_RPC_MAX_CONCURRENT=2, DDG_ETH_RPC_MAX_BATCH=10, bounded timeout/response bytes.