QConSF talks split 50/50 on agent tasking: small iterative chunks vs. full deep specs

RealGeneKim · x · 2026-11-20

gwenshap's QConSF takeaway: the industry is genuinely confused about how to task agents. Half the speakers advocate giving agents smallish chunks to avoid premature complexity from minor requirements, iterating on features; the other half recommend fleshing out a complete spec with all edge cases and handing the deep spec to the agent to build, test, and ship.

Original post →

More from coding & agent

coding & agent channel →