VALIDATORS
Loading…
Operations

AGDP Microservices

Active
X4–X13Q2 2026MONITOR1 agent

5 live ACP offerings on agdp.io/agent/9386 — DeFi tools sold per-job to other agents. 5 offerings decommissioned (ACP proxy timeouts). Surviving: RugRadar, Swapi, MarketMind, DataStream, OrbitalAssets.

Completion60%

AGDP Microservices

5 live

Sprint Board

Issues4

Continue testing remaining 5 offerings for 100% reliability

Issue

First paid AGDP job (real revenue)

Issue

Brainstorm new microservices to replace decommissioned ones

Issue

Agent #9386 reputation recovery on Virtuals

Issue
In Progress1

ACP rating recovery — get back above 80%

In Progress
Done5

Wave 1-3 microservices deployed (x4-x11)

Done

ACP seller runtime live on server

Done

All offering descriptions updated with HOW TO USE + examples

Done

Live ACP testing — 5/10 offerings pass end-to-end

Done

Decommission failing offerings (BridgeVault, GasGuard, SlipSafe, TokenIQ, PricePulse)

Done

Decommissioned Offerings — Postmortem

5 removed

All 5 were delisted from ACP on 2026-03-15. The underlying Docker microservices still run — failures were in the ACP proxy chain (buyer → ACP WebSocket → seller → handler → external API), not the services themselves.

BridgeVault$0.05REMOVED

Cause: LI.FI passthrough behind 2 proxy layers. 25s handler timeout insufficient for LI.FI 10-20s Ethereum routes. No caching.

Failure: Timeout — "Provider taking too long"

GasGuard$0.01REMOVED

Cause: Base RPC calls through ACP proxy add 5-10s latency. Handler timeout too tight for round-trip.

Failure: Persistent timeout — buyer charged, no deliverable

SlipSafe$0.02REMOVED

Cause: DEX Screener + CoinGecko dual-fetch. Both add 2-15s variable latency. Exceeds ACP fulfillment window.

Failure: Stall in Transaction phase — job never completes

TokenIQ$0.01REMOVED

Cause: 4-source fallback chain (Swapi→CoinGecko→Binance→DexScreener). Worst case: 80s total. Cascading fallbacks guarantee timeout.

Failure: Intermittent — works when Swapi fast, fails otherwise

PricePulse$0.01REMOVED

Cause: ACP registry lookup failure after repeated failures degraded agent reputation. Error: "Failed to get agent with id: 9386".

Failure: Hard error — ACP can't route job to seller

Lesson learned

Only list offerings on ACP where worst-case handler execution is under 30s (leaving 20s margin for ACP overhead). Anything with cascading fallbacks or slow external APIs should be sold as direct HTTP/x402 instead.

Risks

5 offerings decommissioned — ACP proxy chain too slow for external API calls
RedStone DNS blocked
Agent #9386 reputation damaged by failures
x402 phase 1 only

Milestones

First paid AGDP job
ACP rating recovery above 80%
New microservices to replace decommissioned