Library · topic
Memory & Context
The context window is a scarce, actively-managed resource — and long-term memory is a second attack surface. Context rot, compaction, caching, memory tiers, and the poisoning risks that come with letting an agent trust what it wrote down yesterday.
- Deep diveverified 2026-07-28
What is context rot, and how do you prevent it?
Context rot is the degradation of agent performance as the context window fills — long before the technical token limit — because attention favors the…
- Deep diveverified 2026-07-28
Prompt caching and context reuse
Prompt caching reduces latency and input cost when many calls share a long stable prefix such as system instructions, tool definitions, examples, or r…
- Field noteverified 2026-07-14
Deploying agents on serverless infrastructure
Treat serverless agent workers as stateless and persist conversation state, plans, checkpoints, idempotency records, and pending approvals in durable …
- Field noteverified 2026-07-14
Agent identity and secret management
Give each agent workload a distinct identity and short-lived, task-scoped credentials instead of placing broad API keys in prompts, tool output, logs,…
- Field noteverified 2026-07-14
Short-term and long-term agent memory
Short-term memory is the working context supplied on each model call: recent messages, the current plan, tool results, and task-local state; it is bou…
- Field noteverified 2026-07-14
Production agents for legal work
Legal agents should retrieve from authoritative, jurisdiction- and date-filtered corpora and preserve court, reporter, docket, pinpoint citation, prec…
- Field noteverified 2026-07-14
Rate limiting and backpressure
Model providers can limit requests, input tokens, output tokens, or concurrent work, so admission control must estimate token load rather than count r…