#agentarchitecture

CLM and turbopuffer's thesis converge: agent context management is being absorbed from both sides.

We have seen banking transform from a physical walk-in through to phone/online web and then to a...

Agent control loops already run classifiers for permissions, routing, and risk. Jev named the pattern. Here is how to build confidence-gated routing.

What an agent loop is and isn't: its three components—state, action, and stopping condition—and why 'build me an app' is input for a chat, not an agent.

My Twitter bot kept drafting replies for 3.5 days after the account was suspended. It had already detected the suspension. Memory was never the problem.

DeepSeek Harness takes a different approach to AI coding agents: instead of locking the model, tools, state, and agent loop into one product, it makes them composable plugins. Here’s how the architecture works, how to run it, and what developers should know before using it.

Six months after an AI agent approves a refund, changes a repository, or produces a board report,...

MCP vs A2A in 2026: 5 tests for whether you need agent-to-agent protocol Summary. The two...

See how Claude-style coding agents really run: harness, loop, permissions, memory, and the minimal prompt that ties them together.

A practical catalog of ai agent control flow patterns that actually ship: bounded retries, idempotent tools, durable checkpoints, human approvals, and debuggable replays.

Agent context compaction drops every block before the summary at 150K tokens. What survives, what instructions silently replaces, and the usage field that lies.

Run agent tool calls in parallel by swapping the loop for a DAG planner: ten round trips become two levels, plus the cap, budget and critic on top.