DevGPT
Cloud-Native AI Coding Agent Platform
An internal platform that enables J.P. Morgan Chase engineers, product & designers to delegate software and prototype development tasks to AI and cloud-based AI agents — from feature implementation to PR review, running securely within the firm's infrastructure at enterprise scale.
Transport only — connection lifecycle, auth, routing, Redis Streams proxying. All harness logic lives exclusively in the worker image. No exceptions.
All agents are stateless. State lives externally in Redis or Temporal's durable execution engine. Temporal's replay guarantees are a first-class design constraint.
Temporal manages intention state continuity across agents and sub-agents — surviving crashes, evictions, failures — while the router/dispatch layer remains stateless.
Our harness is the single source of truth for agent loop logic. IDE plugins are thin context capture and UI bridge layers only. Porting creates divergence risk.
Multiple policies per task are composed with deny-overrides and priority-based tie-breaking. PolicyResolver merges into a single PolicyBundle before agent boot.
Memory Service holds per-user structured operational facts. Brain holds platform-wide institutional narrative. Both queried in parallel and merged at agent boot.
Agents execute untrusted code in sandboxed subprocesses using Anthropic's Sandbox Runtime, which leverages bubblewrap for lightweight, kernel-level process isolation — no privileged containers, no separate runtime service. Leveraging Istio Integrated directly into the DevGPT plugin and CLI. Sub-2-second dispatch target with warm pod pool strategies via Karpenter NodePool and KEDA ScaledObject.