Your AI Agent Forgets Your Rules After Hour One: The Context-Hygiene Check

Description and use

<p>You set up an AI agent with clear rules: "Always cc my assistant on external emails," "Never delete files without asking," "Flag invoices over $500 before paying." It follows them perfectly for the first hour. Then it starts drifting. By hour three, it's drafting replies without the cc line and moving files you didn't tell it to touch.</p> <p>The agent isn't being disobedient. The rules just got buried.</p> <h2> Why agents forget </h2> <p>AI agents process information in a context window — essentially a rolling transcript of everything they've seen and done. Research from Chroma, a vector-d

Problem

<p>You set up an AI agent with clear rules: "Always cc my assistant on external emails," "Never delete files without asking," "Flag invoices over $500 before paying." It follows them perfectly for the first hour. Then it starts drifting. By hour three, it's drafting replies without the cc line and moving files you didn't tell it to touch.</p> <p>The agent isn't being disobedient. The rules just got buried.</p> <h2> Why agents forget </h2> <p>AI agents process information in a context window — essentially a rolling transcript of everything they've seen and done. Research from Chroma, a vector-d

Outcome

<p>You set up an AI agent with clear rules: "Always cc my assistant on external emails," "Never delete files without asking," "Flag invoices over $500 before paying." It follows them perfectly for the first hour. Then it starts drifting. By hour three, it's drafting replies without the cc line and moving files you didn't tell it to touch.</p> <p>The agent isn't being disobedient. The rules just got buried.</p> <h2> Why agents forget </h2> <p>AI agents process information in a context window — essentially a rolling transcript of everything they've seen and done. Research from Chroma, a vector-d

Evidence

Your AI Agent Forgets Your Rules After Hour One: The Context-Hygiene Check

Source