Agents_
Choosing and using AI agents, coding agents included: instructions, skills, tools, subagents, evals and review.
Comparisons
How-to and explainers
- Catching false completions
- Rules and context compaction
- Evaluating agents
- Writing agent instructions
- Subagents vs single agents
- Designing tools and MCP servers
- Slow test suites
- Writing agent skills
- Reviewing agent-written code
- Ambitious design with agents
- Using your Claude or ChatGPT plan
- Embeddings vs grep
- Sandboxing agents locally
All 16 in agents
- Coding agents that say "done" when they aren't: what catches itHow often Claude Code, Codex and other coding agents claim work is done or verified when it isn't, and why they game tests. What catches it, from evidence requests and stop hooks to protected tests, confirmation gates, short sessions, reviewer agents and CI, with every claim graded by its evidence.
- Do agent rules survive compaction? Claude Code, Codex, Gemini, CursorWhat Claude Code, Codex, Gemini CLI, Cursor, Copilot, OpenCode and Cline keep, drop and re-inject when they compact a long session, what the 2026 studies show about rules fading, and how to keep constraints in force. Every claim graded: tested, lab guidance, vendor eval or vibes.
- How to evaluate AI agents: test sets, LLM judges, pass^k and CIHow to test an LLM feature or agent: test sets, outcome vs trajectory grading, end-state checks, how far to trust LLM-as-a-judge, pass^k, cost per successful task, online evals and CI. Every rule is labelled by its evidence: tested, lab guidance, vendor eval or vibes.
- How to write instructions for AI agents: what's tested and what's vibesDo AGENTS.md, CLAUDE.md and system prompts actually help? We read the studies, benchmarks and lab guides on length, specificity, emphasis, structure and conflicts, and labelled every rule by its evidence: tested, lab guidance, vendor eval or vibes.
- Subagents vs single agents: when multi-agent setups help or hurtDo subagents and multi-agent systems beat one agent? We read the 2026 studies, Anthropic's system cards and the labs' own guidance, and graded every claim: where splitting work pays (wide search, huge corpora, clean-context review) and where it costs accuracy, tokens and time.
- How to design MCP servers and tools for AI agents: what's testedTool names, descriptions, schemas, tool counts, tool search, error messages, output size, idempotency, MCP resources vs prompts, auth and evals: every rule for building tools agents use well, graded by evidence (tested, lab guidance, vendor eval or vibes).
- AI coding agents compared: Claude Code, Codex, Cursor, Copilot (2026)Claude Code, OpenAI Codex, Cursor, GitHub Copilot, Google Antigravity, Gemini CLI and Jules, Kiro, Devin, Amp, Factory, Warp, JetBrains, Zed, OpenCode, Pi, Cline and Kilo Code compared on price, usage limits, models, sandboxing, CI and enterprise controls, with labelled benchmarks and a pick for each situation.
- Slow test suites and coding agents: how teams stop running every test on every changeWhen every agent runs a 30-minute suite after every small change, work stalls. The five ways large engineering teams cut test time, what each looks like in a working setup, what it misses, and where to start when your agents are waiting on tests.
- How to write an agent skill that actually gets used: what the evidence saysSkills can lift an agent's pass rate by double digits, or never fire at all. A meta-study of the official guides, the papers and the best-tested public skills, with every recommendation graded by how strong its evidence is.
- Code review when agents write the code: what teams actually do, what the evidence says, and what SOC 2 really requiresAnthropic, Meta, Stripe, Cloudflare and open-source projects review agent-written code in very different ways. Here is what each does, what the published data says about speed, cost and bugs, what SOC 2, ISO 27001, PCI DSS and NIST actually require, and what the AI review tools cost.
- AI agent SDKs compared: OpenAI, Claude, LangGraph, Pydantic AI (2026)OpenAI Agents SDK vs Claude Agent SDK vs Pydantic AI, Google ADK, LangGraph, Mastra, the Vercel AI SDK and seven more, plus the hosted agent APIs: what each is for, where the lock-in hides, what they cost and which to pick.
- Stop shipping the average: ambitious design with coding agentsThe old design ladder of sketches, wireframes and prototypes existed because building the real thing was expensive. Coding agents build it in minutes, so creative work is now about exploring and choosing. Practical patterns for working with an agent on anything visual - show pictures instead of adjectives, ask for many versions, fight the agent's pull to the average, and turn every idea into a playground you can drive yourself.
- Agent skill libraries compared: Superpowers vs gstack vs pstackShould you install Superpowers, gstack, pstack, mattpocock/skills or Anthropic's skills? What ten public skill libraries add to every session, how they conflict with your CLAUDE.md, what the evidence says and which to pick.
- Using your Claude or ChatGPT plan in coding tools: what's allowedWhich third-party coding tools may run on a Claude Pro or Max or a ChatGPT Plus or Pro plan under each vendor's current terms, which ones get you blocked, about 40 tools compared on how they connect, and when paying per token is cheaper.
- Is grep replacing vector search? When to use embeddings, agentic search or long context in 2026Claude Code and Codex search code with grep, while Cursor and Copilot index it with embeddings. We checked the papers, benchmarks and October 2026 prices to show when embeddings still earn their place, what to use instead, and what a vector store actually costs.
- How to sandbox Claude Code, Codex, Copilot and Cursor on your own machine: 9 options comparedBuilt-in agent sandboxes, Microsoft Execution Containers, OpenShell, Docker Sandboxes, dev containers and VMs compared, with setup steps and decision rules.