After 400+ production AI agents: the 7 skills every agent engineer needs
MaryamMiradi · x · 2026-11-19
Drawing on 400+ production AI agents, the author lists 7 skills for agent engineers:
- System design: an agent is an orchestration of LLM, tools, DBs, sub-agents — with failure isolation, not spaghetti.
- Tool & contract design: strict types, examples, input validation; treat tool schemas like legal contracts.
- Retrieval engineering: chunking, hybrid search, reranking — bad context beats even the best LLM.
- Reliability engineering: retries, timeouts, fallbacks, circuit breakers — the backend playbook.
- Security & safety: validate inputs/outputs, least-privilege tools.
- Evaluation & observability: trace every tool call, regression-test against known results.
- Product thinking: show what the agent did, surface uncertainty, keep humans in the loop.
Takeaway: frameworks change, systems thinking doesn't.
More from coding & agent
- Devs flag GPT's defensive programming for hiding bugs; OpenAI asks for /feedback cases — pvncher · 2026-11-19
- Flutter ships A2UI 1.0: agents respond with real interactive widgets, not chat text — rseroter · 2026-11-19
- Skills vs MCP vs RAG vs Memory: the four context mechanisms AI agents actually need — MaryamMiradi · 2026-11-19
- Anthropic details agent memory: 'dreaming' lets Claude organize its experience — daniel_mac8 · 2026-11-19
- The case for externalized memory as the real fix for continual learning — daniel_mac8 · 2026-11-19
- FT: US tech outages blamed on rush to fix bugs found by AI — nordicinst · 2026-11-19