What MCP debates on Reddit and HN actually argue: context bloat, redundant servers, security boundaries
Fine_Airline_6832 · reddit · 2026-11-02
A write-up distills the recurring MCP debates across Reddit and Hacker News into three arguments:
- Schemas spend the window: MCP tool definitions can take 1/3 of context on every message whether used or not; some report 30-50% loss on fresh sessions, with disabled integrations still injecting instructions. Check with /mcp and /context.
- Many servers are redundant: models already know gh and similar commands — the GitHub MCP server 'consumes context and adds nothing'.
- The boundary case still holds (simonw): an agent with shell and internet can call APIs directly; MCP earns its place when you control the services, the model never sees the API key, and there's a log. Operational gaps remain: no standard trace across servers, refresh tokens outliving sessions.
Takeaway: a sentence in the prompt is not a permission, and a leftover server is a bill paid before the first word.
More from coding & agent
- Reward hacking is Goodhart's Law for agents, warns Morgan Brown — morganb · 2026-11-02
- Debate on agentic commerce monetization: do agents really need payments revenue? — jeff_weinstein · 2026-11-02
- Veynod: open-source agent coding runtime with one model loop and a deterministic kernel — TKOWNS · 2026-11-02
- Claude Code subagents silently switch to paid usage, burning $100 in credits — nptacek · 2026-11-02
- Instinct reportedly blocked from WhatsApp as observers say agent pivot was predictable — itsOmSarraf_ · 2026-11-02
- Purser: an MCP server that lets agents pay with smart-contract limits a jailbreak can't bypass — Plane-Preference6245 · 2026-11-02