Graph Engineering and Harness Engineering: a blueprint for building reliable agentic systems
Pavan_Belagatti · x · 2026-10-12
Developer Pavan Belagatti outlines two engineering practices for building reliable agentic systems, with an accompanying diagram.
Graph Engineering
- Instead of one massive prompt, design work as a graph of smaller explicit tasks—each with a purpose, defined input, output, and routing to the next step.
- Benefits: structure, inspectable decisions, and parallelism. Key elements include nodes/agents, edges/routing, dependencies, state & memory, human approval, and observability.
Harness Engineering
- Code generation is only a small part of software engineering; reliable agents need an operating environment around the model: tools, memory, context engineering, sandboxing, orchestration, and serving layers.
- Multi-agent workflows (metrics analysis, log search, deployment checks, root-cause correlation) must be organized in the right order with shared context.
More from coding & agent
- LangChain Founder: Persistent Machines Are a Weakness — Agents Run Better Fully Parallel — zeeg · 2026-10-12
- MCP servers eat 29k tokens before you type a word — dev ships mcp-tax CLI to measure the tax — princejain756 · 2026-10-12
- Opus 5.5 builds its own SVG drawing tool, then drafts a detailed PS5 controller cutaway — ImportGuy · 2026-10-12
- Cross-harness agent messaging: one Redditor's file-based bus spiraled into thousands of messages — positiveconstraint · 2026-10-12
- GPT-6 Astra nails 72-hour agent tasks with 63% on-time rate, crushing Fable 5.1's 4% — maksym_andr · 2026-10-12
- Voice-controlling your coding agent from iOS and resuming sessions at your desk: the always-on Mac setup debate — TheZachMueller · 2026-10-12