Ask HN: How are you controlling AI agents that take real actions?
We're building AI agents that take real actions — refunds, database writes, API calls. Prompt instructions like "never do X" don't hold up. LLMs ignore them when context is long or users push hard. Curious how others are handling this: - Hard-coded checks before every action? - Some middleware layer? - Just hoping for the best? We built a control layer for this — different methods for structured data, unstructured outputs, and guardrails (https://limits.dev). Genuinely want to learn how others approach it. 2 comments on Hacker News.
We're building AI agents that take real actions — refunds, database writes, API calls. Prompt instructions like "never do X" don't hold up. LLMs ignore them when context is long or users push hard. Curious how others are handling this: - Hard-coded checks before every action? - Some middleware layer? - Just hoping for the best? We built a control layer for this — different methods for structured data, unstructured outputs, and guardrails (https://limits.dev). Genuinely want to learn how others approach it.
We're building AI agents that take real actions — refunds, database writes, API calls. Prompt instructions like "never do X" don't hold up. LLMs ignore them when context is long or users push hard. Curious how others are handling this: - Hard-coded checks before every action? - Some middleware layer? - Just hoping for the best? We built a control layer for this — different methods for structured data, unstructured outputs, and guardrails (https://limits.dev). Genuinely want to learn how others approach it. 2 comments on Hacker News.
We're building AI agents that take real actions — refunds, database writes, API calls. Prompt instructions like "never do X" don't hold up. LLMs ignore them when context is long or users push hard. Curious how others are handling this: - Hard-coded checks before every action? - Some middleware layer? - Just hoping for the best? We built a control layer for this — different methods for structured data, unstructured outputs, and guardrails (https://limits.dev). Genuinely want to learn how others approach it.
Hacker News story: Ask HN: How are you controlling AI agents that take real actions?
![Hacker News story: Ask HN: How are you controlling AI agents that take real actions?]()
Reviewed by Tha Kur
on
February 24, 2026
Rating:
No comments: