Pick the right models. Give them the right infrastructure

Описание и применение

Агенты забывают, и их память не прозрачна. Внедрена файловая система памяти с метками, лимитами и сканированием секретов, обеспечивающая аудит и надёжность.

Задача

Agents forget and their memory is opaque, making it hard to audit or trust what they remember.

Результат

Agents now have auditable, size-limited, and time-bound memory, improving reliability and transparency.

Рабочий процесс

• 1. Identify memory issues in agents. 2. Design plain-file memory with receipts. 3. Enforce size thresholds and age-out rules. 4. Run a secrets scan before data ingestion. 5. Publish rules and audit logs publicly. 6. Iterate with new rules and measurements.

Доказательства

Agents forget. And when they do remember, you usually can't see what they remember, where it came from, or why it is still there.

Источник

We built the discipline we needed: memory as plain files, with rules. Every memory carries a receipt — a date and a source.

Источник

Memory files have size thresholds and age out when idle. A secrets scan runs before anything lands.

Источник