Reading

What I'm reading.

Notes and summaries of agent-engineering work by others — not my own writing. The pieces that shaped how I think about agents, grouped by theme.

Orchestration

Building effective agents Anthropic · 2024-12-19

The canonical starting point — workflow vs agent, and five composable orchestration patterns.

A practical guide to building agents OpenAI · 2025

When to reach for an agent, single→multi-agent evolution, and guardrails at every stage.

How we built our multi-agent research system Anthropic · 2025-06-13

Orchestrator-worker in production; ~15× the tokens, so reserve it for high-value work.

Building a C compiler with a team of parallel Claudes Anthropic · 2026-02-05

16 parallel Claudes over ~2,000 sessions wrote a 100k-line C compiler that builds the Linux kernel — agent teams as a supervision pattern.

Context

Effective context engineering for AI agents Anthropic · 2025-09-29

Context is a finite attention budget; the job is curating what goes in.

Context engineering for AI agents: lessons from building Manus Manus · 2025-07-18

KV-cache hit rate, the filesystem as external memory, and keeping failures in context.

Tools

Writing effective tools for agents Anthropic · 2025-09-11

Evaluation-driven tool design; fewer, higher-leverage tools beat many overlapping ones.

Advanced tool use Anthropic · 2025-11-24

Tool search, programmatic tool calling, and tool-use examples — big token and accuracy wins.

The "think" tool Anthropic · 2025-03-20

A pause-and-reason step for tool-heavy, policy-dense decision chains.

New tools for building agents OpenAI · 2025-03-11

Responses API + open-source Agents SDK; the Agents and Handoffs primitives.

Skills

Equipping agents for the real world with Agent Skills Anthropic · 2025-10-16

SKILL.md folders + progressive disclosure, shipped as an open standard.

Manus embraces open standards: Agent Skills Manus · 2026-01-27

Adopting the Skills standard; Skills and MCP are complementary, not competing.

Introducing Project Skills Manus · 2026-02-12

Curate a per-project skill library; turn team expertise into a reusable asset.

Harness

Effective harnesses for long-running agents Anthropic · 2025-11-26

Init-agent + coding-agent, a JSON feature list, and context recovery across windows.

Introducing Manus Projects Manus · 2025-12-01

Persistent project workspaces so setup becomes reusable, not repeated.

Projects that learn from every task Manus · 2026-05-06

Propose-and-approve self-updating instructions and skills.

Harness design for long-running application development Anthropic · 2026-03-24

Planner–generator–evaluator: splitting generation from evaluation beats self-review — then a 1M-token window killed the sprint structure entirely.

Evals

Demystifying evals for AI agents Anthropic · 2026-01

Automated evals pre-launch, monitoring post-launch, A/B once there is traffic; start from 20–50 tasks drawn from real failures.

Safety

How we built Claude Code auto mode Anthropic · 2026-03

Classifiers instead of permission prompts — an injection probe on inputs, a transcript classifier on actions, with honest FP/FN numbers.

How we contain Claude across products Anthropic · 2026-05-28

Supervise what the agent can do, not what it does — sandboxes, VMs, egress control; users approve 93% of prompts and stop reading them.

Platform

Introducing AgentKit OpenAI · 2025-10-06

Agent Builder, Connector Registry, ChatKit, and expanded evals.

The next evolution of the Agents SDK OpenAI · 2026-04-15

From pure orchestrator to a full execution environment — native sandbox, harness, memory.

Scaling Managed Agents: Decoupling the brain from the hands Anthropic · 2026-04-08

A hosted service for long-horizon agent work — interfaces stay stable while the harness underneath keeps changing.

How agents are transforming work OpenAI · 2026-06-28

Field data on delegation: 80% of requests map to 30+ minutes of human work — the horizon is lengthening fast.