Coding Agents in Practice (2/5) — Cursor 3 vs Claude Code vs GitHub Copilot (May 2026)

AI ์ฝ”๋”ฉ ์—์ด์ „ํŠธ ์‹ค์ „ (2/5) — Cursor 3 vs Claude Code vs GitHub Copilot 2026-05 ๋น„๊ต

"Which one is better?" is the wrong framing. "Which one fits my workflow?" is the real question.

ํ•ต์‹ฌ ์š”์•ฝ

  • Three coding agents, three philosophies: Cursor 3 (IDE-first), Claude Code (CLI-first), Copilot (editor inline)
  • Pricing, subscription model, and four capability axes laid side by side. All prices verified 2026-05-05 from official pages
  • Conclusion: same category, but they are not 1-to-1 substitutes — design philosophy is the deciding factor
  • Subscription decision reduces to: "Do I live in the IDE, in the terminal/workflow, or in inline autocomplete?"

1. Design philosophy of the three tools

Tool Where it lives Primary mode Model
Cursor 3 Standalone IDE (VS Code fork) Chat + Composer + cloud Background Agent OpenAI / Anthropic / Cursor's own
Claude Code Terminal / IDE extensions CLI workflows + skills + hooks Anthropic Opus / Sonnet / Haiku
GitHub Copilot VS Code, JetBrains, etc. Inline autocomplete + Chat + Agent Mode OpenAI GPT-family + optional models

Same "AI coding," different workspace. Cursor's workspace is the IDE; Claude Code's is the workflow; Copilot's is the file you're editing.

2. Pricing comparison (verified 2026-05-05)

Prices change. Re-verify on official pages at decision time.

Plan Cursor 3 Claude Code GitHub Copilot
Free / trial 2-week Pro trial + limited free Limited free with Anthropic account Free for students / OSS maintainers
Pro (individual) $20/month Usage-based ($20+/month typical) $10/month (Individual)
Team / Business $40/month/user (Business) Usage-based $19/month/user (Business)
Top tier $200/month (Ultra) Usage-based $39/month/user (Enterprise)

Pricing model differences: - Fixed subscription (Cursor / Copilot): predictable monthly cost. Heavier users get better unit economics. - Usage-based (Claude Code): per-token cost × actual usage. Light users pay less, heavy users pay more — less predictable, but matches actual consumption.

Decision rule: more than ~1 hour of agent use per day → fixed subscription. Occasional use → usage-based.

3. Four capability axes

3.1 Context management

Capability Cursor 3 Claude Code Copilot
Auto codebase indexing ✅ (whole project) Partial (read on request) ✅ (current workspace)
Explicit context rules .cursorrules CLAUDE.md .github/copilot-instructions.md
Memory / cross-session persistence Partial (per chat) Strong (3-tier memory + session recovery) Weak (per session)

Claude Code's memory and session-recovery story is the most mature among the three (see part 1/5).

3.2 Automation / multi-agent

Capability Cursor 3 Claude Code Copilot
Background / async agent ✅ Background Agent (cloud VM) ✅ Subagents + background bash ๐Ÿ”ถ Agent Mode (beta)
Hooks (event-driven execution) ❌ ✅ (PreToolUse, PostToolUse, SessionStart, etc.) ❌
MCP integration ✅ ✅ ๐Ÿ”ถ Partial

Claude Code's hooks give it the deepest automation surface. Cursor 3's cloud-isolated VM is the strongest fit for heavy builds and refactors.

3.3 Model freedom

Capability Cursor 3 Claude Code Copilot
Model switching ✅ (OpenAI / Anthropic / Cursor) Anthropic only (Opus / Sonnet / Haiku) ✅ (GPT-family + add-ons)
Local model support Partial ❌ ❌

If multi-provider routing matters, Cursor 3 wins. If you're going deep inside the Anthropic ecosystem, Claude Code wins.

3.4 IDE / editor integration

Capability Cursor 3 Claude Code Copilot
Inline autocomplete ✅ ❌ (CLI-first) ⭐ (strongest)
Multi-file edits (Composer) ✅ ✅ Partial
Editor depth Own IDE VS Code extension / terminal VS Code, JetBrains, etc.

For natural inline autocomplete while editing, Copilot is the most polished.

4. Decision rules

Profile Recommendation
IDE-resident; inline autocomplete and chat are the core Cursor 3 or Copilot
Terminal / workflow / automation is the core; CLAUDE.md / skills / hooks matter Claude Code
≤1 hour/day of light help Copilot Individual ($10/month)
Heavy background tasks (builds, refactors, test automation) Cursor 3 Ultra ($200/month) or Claude Code + hooks
Multi-model routing (OpenAI + Anthropic + custom) Cursor 3
Anthropic ecosystem depth (memory, sessions, MCP) Claude Code

Stacking all three is not unusual: Copilot for inline autocomplete, Cursor 3 for IDE chat / Composer, Claude Code for workflow and automation. But cost stacks too — most teams pick one primary workspace and pay for it, then keep one or two as supplements.

5. At a glance

Axis Cursor 3 Claude Code Copilot
Workspace IDE Terminal / workflow Editor inline
Automation depth Medium-High (Background Agent) High (hooks + skills) Medium (Agent Mode beta)
Pricing $20–$200/month Usage-based $10–$39/month
Model freedom High (multi-provider) Anthropic only OpenAI + add-ons
Memory / session recovery Medium High Low

The "which is better?" answer is always "depends on the workflow." Look at the axis that maps to your day, and pick the one that hurts least there.

Next up

Part 3/5: Building MCP Servers — Spec, Examples, Debugging. All three tools support the Model Context Protocol; this entry walks through writing an MCP server from scratch.

References

  • Anthropic, Claude Code Pricing — claude.com/pricing (verified 2026-05-05).
  • Cursor, Pricing — cursor.com/pricing (verified 2026-05-05).
  • GitHub, Copilot Plans — github.com/features/copilot/plans (verified 2026-05-05).
  • Part 1/5 — Claude Code Workflow.

This is part 2/5 of the Coding Agents in Practice series.

๋Œ“๊ธ€

๋Œ“๊ธ€ ์“ฐ๊ธฐ

์ด ๋ธ”๋กœ๊ทธ์˜ ์ธ๊ธฐ ๊ฒŒ์‹œ๋ฌผ

Agent Memory Engine (2/10) — Building an AI Agent Memory System with SQLite Alone

← 1/10 SQLite Memory Engine f… ๐Ÿ“š Series Index 3/10 memcore Completed → A memory engine that runs without daemons, without dependencies, anywhere ํ•ต์‹ฌ ์š”์•ฝ This post covers how to design an AI agent memory engine (memcore) on a single-file SQLite backend. - Structure: 25 modules, ~6,464 lines, 22 DB tables - Principles: no daemon, single-file portability, no required external dependencies, host-agnostic - Core techniques: 4-layer hybrid search (topic · vector · FTS5 · LIKE), U-tag dialectic-based confidence evolution, ingestion-layer bias rejection rules What This Post Covers One approach to building AI agent memory without a server or cloud vector DB. Specifically: (1) the limitations of text-file-based memory, (2) vector DB alternatives compared and selection criteria, (3) how to layer search tiers, (4) how to evolve confidence scores over time, and (5) how to block biased memory accumulation at the ingestion stage. Design Principles: Why Single-File SQLite Text-file-based ...
Read more »

"ML Foundations (9/9) — PyTorch vs TensorFlow, and the Road to Local LLMs"

์ด๋ฏธ์ง€
← 8/9 Deep Learning Architec… ๐Ÿ“š Series Index (series end) The final part. So far we've focused on models. Now we focus on the tools that actually run them . We'll lay out the philosophical difference between PyTorch and TensorFlow, and trace how Hugging Face transformers , llama.cpp , MLX , and Ollama built the bridge to running large language models on your own machine. By the end you should have the full mental model of "download a pretrained LLM and serve it locally." 0. Learning Objectives Explain the eager-vs-graph difference between PyTorch and TensorFlow. Explain, in graph terms, how autograd automates the backward pass. Use Hugging Face transformers ' AutoModel , AutoTokenizer , and pipeline abstractions. Describe llama.cpp's GGUF quantization, INT4 inference, and CPU-first flow. Describe Apple MLX's use of unified memory and how it differs from PyTorch. Run a local LLM with Ollama and call it through an OpenAI-compatible API. ...
Read more »

"RAG Core Study (14/26) — Evaluation Sets with RAGAS & DeepEval"

์ด๋ฏธ์ง€
Series overview: Series index ๐Ÿ“š Series Index Without an evaluation set, every RAG improvement is a story, not evidence. RAG systems fail in more than one place: retrieval, context selection, answer generation, and grounding. That means "looks better" is not a metric. Part 14 explains how to build a golden dataset , how RAGAS and DeepEval fit into the workflow, and why teams must separate retrieval quality from answer quality if they want tuning decisions to hold up. 0. Prerequisites Part 13 reranking — system quality now depends on multi-stage ranking. Part 1 grounding — a correct-looking answer is not enough. Part 12 Hybrid — multiple retrieval settings need comparison on fixed data. 1. Learning Objectives Build a small but useful golden dataset for RAG. Distinguish retrieval metrics from answer metrics . Use RAGAS and DeepEval in the right roles. Avoid the common traps in judge-model-based evaluation. 2. ํ•ต์‹ฌ ์š”์•ฝ An evaluation set for RAG is ...
Read more »

"ML Foundations (8/9) — Deep Learning Architectures: CNN, RNN, Attention"

์ด๋ฏธ์ง€
← 7/9 Deep Learning Training ๐Ÿ“š Series Index 9/9 PyTorch vs TensorFlow → What did we keep adding on top of an MLP? Vision went CNN, sequences went RNN/LSTM, and eventually both converged on Attention and the Transformer. Each architecture is easier to remember if you read it as what it gave up to gain something else . This part is the one-line summary of the decisive inflection points: LeNet → AlexNet → ResNet → LSTM → Attention → Transformer. 0. Learning Objectives Explain why CNN's convolution, pooling, and weight sharing match images so well. Trace what got unblocked from LeNet → AlexNet → VGG → ResNet as depth grew. State BPTT for RNNs and the vanishing/exploding gradient problem. Write the LSTM gate equations (forget, input, output) with intuition. Write the attention formula in query/key/value form and the scaled dot-product variant. State the precise reasons Transformers replaced RNNs (parallelism + long-range dependencies). 1. ํ•ต์‹ฌ ์š”์•ฝ CNN : weight sharin...
Read more »

"ML Foundations (7/9) — Deep Learning Training: Optimizers, Regularization, Initialization"

← 6/9 Neural Networks ๐Ÿ“š Series Index 8/9 Deep Learning Architec… → Building an MLP, as we did in Part 6, is not the same as getting it to train well . The right optimizer, regularization, initialization, and learning rate — when these line up, deep networks converge. When they don't, the network refuses to learn at all. This part is the catalog of those crafts, with formulas, paper citations, and code in one place. 0. Learning Objectives Compare and write the update rules for SGD, Momentum, Nesterov, and Adam. Explain how Dropout, BatchNorm, and LayerNorm work and where they belong in a model. Derive the variance formulas for Xavier and He initialization and match them to activations. Implement step, cosine, and warmup learning-rate schedules in PyTorch. Explain why gradient clipping is effectively required for RNNs and Transformers. Diagnose the most common training failures (NaN, plateau, overfitting) and apply first-line fixes. 1. ํ•ต์‹ฌ ์š”์•ฝ SGD : \(w \leftarro...
Read more »

OpenClaw to Hermes Migration (2/13) — What to Preserve, Partially Port, or Discard

← 1/13 Current Structure Inve… ๐Ÿ“š Series Index 3/13 What Moves to Hermes → A code review record classifying operational scripts into three tiers: preserve, partial-preserve, and discard. What This Post Covers An asset classification framework (preserve / partial-preserve / discard) for migrating a legacy agent system to a successor A concrete case applying the Strangler Fig + Subset Migration strategy to a memory pipeline Classification results evaluated across LOC, external dependencies, and replaceability axes A method for slimming down a bloated single file (1,310 LOC) using the migration as a natural entry point Problem Definition: Port Everything or Cut Deep? The OpenClaw inventory comprises multiple agents, multiple scheduler/daemon processes, and multiple operational scripts. When moving to the successor system, Hermes, the choice is not binary. A full port carries accumulated technical debt along with it; a minimal port discards hard-won domain logic. Code-revie...
Read more »

AI Agents I Built (5/7) — Building an Automated Blogger API Publishing System

← 4/7 My Life Organizer Agent ๐Ÿ“š Series Index 6/7 Dual → Designing a Markdown-Based Auto-Publish Pipeline with Blogger API v3 + OAuth 2.0 Key Summary Blogger API v3 + OAuth 2.0 enables a pipeline that converts Markdown files into publishable HTML posts blogger_publish.py handles the full flow — frontmatter parsing → body cleanup → HTML conversion → CSS injection → API post — in a single script For bulk publishing, quota management (delay, retry, TOC suppression) is the critical design concern Platform Selection Criteria API access is mandatory for AI agent-driven content publishing. A comparison of platforms: Platform API Cost AdSense Decision WordPress.com REST API Paid plan required Paid only Cost overhead Ghost REST API Self-hosted or paid Manual setup Ops overhead Blogger v3 API Free Native integration Adopted Blogger is free, exposes a REST API, and integrates natively with AdSense. For personal blog automation, this combination i...
Read more »