# \#agents

**URL:** https://siai.forum/tag/agents/23.md

[Latest](https://siai.forum/latest.md) · [Categories](https://siai.forum/categories.md) · [Tags](https://siai.forum/tags.md)

---

## [MCP and tool-use standards: are you adopting them, and how?](https://siai.forum/t/mcp-and-tool-use-standards-are-you-adopting-them-and-how/37)

<div class="topic-metadata">

**Author:** [@admin](https://siai.forum/u/admin)\
**Replies:** 0\
**Last updated:** [October 16, 2026, 7:05pm UTC](https://siai.forum/t/mcp-and-tool-use-standards-are-you-adopting-them-and-how/37 "2026-10-16T19:05:30Z")

</div>

Standard protocols for connecting models to tools and data, such as the Model Context Protocol, are spreading across clients and frameworks. Are you building or using MCP servers? For what? How are you handling auth, p…

---

## [Agents in production: what actually works and what breaks?](https://siai.forum/t/agents-in-production-what-actually-works-and-what-breaks/36)

<div class="topic-metadata">

**Author:** [@admin](https://siai.forum/u/admin)\
**Replies:** 0\
**Last updated:** [October 16, 2026, 7:05pm UTC](https://siai.forum/t/agents-in-production-what-actually-works-and-what-breaks/36 "2026-10-16T19:05:30Z")

</div>

Agent demos are easy; agents that run reliably for real users are hard. If you’ve shipped one: What’s the task, and how autonomous is it? Where does it fail most: planning, tool calls, long context, cost, or evaluatio…
