Agents

Explains function calling (tool use) in LLMs: how models emit structured requests to invoke external functions, the request-execute-return loop, provider support, and practical reliability notes.

As local LLMs improve, harnesses are learning to pair them with frontier models. A look at the four collaboration patterns already shipping in open source.

Distinguishes RLVR as training-time weight updates from inference-time agent verification loops.

Paper summary of AdapTime, an adaptive planner for temporal reasoning in LLMs.

Explains graph-based memory for LLM agents, including taxonomy, GAM consolidation, and hybrid retrieval.

Principles for designing AI harnesses: context, tools, verification, autonomy, observability, and composition.

Explains ACP, a JSON-RPC protocol connecting editors to AI coding agents with client-brokered tools.

Pattern where agents declare structured pre-tool intent for gating, auditability, and drift detection.

Distinguishes sub-agents from tool-agents by autonomy, interface, context, and harness contract.

Concrete web-search example showing how Tool-DC strategic anchor grouping reduces schema-confusion in tool calls.

Overview of AgentFlow, an agent architecture that trains a planner with Flow-GRPO for multi-turn tool use.

Overview of Tool-DC, a try-check-retry framework for robust long-context tool-calling with large tool registries.

Verification checkpoint in agent harnesses that blocks irreversible actions until cross-skill, cross-scale, and evidence-sufficiency checks pass.

Overview of agent harness engineering — the scaffolding, infrastructure, and tooling surrounding an AI agent, covering execution environments, tool orchestration, memory management, control flow, tracing, safety, and state persistence