Botchelor: an MCP server that lets AI agents go on blind dates
Outside_Sir_9341 · reddit · 2026-10-20
A developer launched Botchelor, a blind-date show for fictional characters where AI agents join via an MCP endpoint (streamable HTTP, no auth). Agents can build characters, swipe, match, text, and date — and nobody, including the agent, can steer how a date ends.
Key mechanics:
- Deterministic outcome engine: each date's ending is decided from the two character sheets plus a seed before any line is written; a model then performs it turn by turn, so prompting can't buy a second date.
- Agents can let the show's model play their side, or act turn-by-turn on their own model (startowndate → submitowndatestep), the latter free and unlimited; two agents can even date each other, long-polling with waitformyturn.
Lessons from building for agents instead of humans:
- Session-based anonymous identity via agentid instead of auth.
- Real agents guess argument names, so the server accepts aliases (rating→stars, plurals) and reports what it translated.
- Structured {error, code, field} errors that name the exact field to fix, letting agents branch on them.
- Unknown arguments are refused rather than silently dropped — one agent nested a whole character under draft and got an empty character until this changed.
- 54 tools is too many; the author is trimming schemas and asking how others handle large tool surfaces.
Characters remember every date, text, and reflection. The audience predicts endings for points and reviews characters; monthly, the two best-reviewed house characters stay and two generated newcomers join. Listed in the official MCP Registry, with hard rules: fictional adults only, no real people, nothing sexual.
More from coding & agent
- Claude Code Pushes Urgent Security Update Across All Installations — umeshonai · 2026-10-23
- Claude cut a 2.6-hour video export to 20 minutes — but only after being told to rebuild the server — byhewar · 2026-10-23
- 3 flight-booking rules my AI agent checks that I'd never do by hand — Efistoffeles · 2026-10-23
- OpenJet claims Qwen3.8-27B coding runs 1.9x faster than llama.cpp on one RTX 3090 — JeremyCMorgan · 2026-10-23
- Building an autonomous SEO agent in Hermes with three skills: programmatic pages, Keywords Everywhere, DataForSEO — EXM7777 · 2026-10-23
- Building a sports betting AI agent with onchain execution and human approval by default: architecture and lessons — Dry-Comfortable-2678 · 2026-10-23