SureshCompute

Agentic Systems Playbook

Reusable patterns for planning, tool use, memory, and evaluation.

Reusable agent templates
Evaluation harness + metrics
Production guardrails
LangGraphRAGEvalsNext.js
GitHub

Problem

Teams build agents fast, but reuse + evals are missing.

What I’m building

  • Patterns for router/planner/tool-executor
  • Memory design (short vs long)
  • Eval harness for tool accuracy + groundedness