MCP deferred tools explained: progressive disclosure, like a CLI, model calls first to fetch params
eherrerosj · x · 2026-09-16
A discussion of how MCP's deferred tools work: it's essentially progressive disclosure — like a CLI that only reveals full params when invoked, the model must call the tool first to fetch its complete parameter definitions, avoiding stuffing every tool description into the context upfront.
More from coding & agent
- The coordination tax: why parallel multi-agent lanes burn tokens on excessive verification — pvncher · 2026-09-16
- Prior Labs releases TabPFN-3.5, a tabular foundation model handling 1M rows, tops benchmarks — tuanacelik · 2026-09-16
- Telling your coding agent to slow down actually helps, developer reports — nilogram · 2026-09-16
- The Hacker's Guide to Attacking AI Agents: a systematic look at agent attack surfaces — _clickfix_ · 2026-09-16
- vAIvar: open-source honeypot feeds attacking AI agents endless hallucinations — Scobleizer · 2026-09-16
- Polyphonic details: auto-imported skill libraries, tag-a-into-chat agents, early Mnemos memory system — RileyRalmuto · 2026-09-16