#agentic-engineering

[anupkaranjkar]

The Death of Traditional Prompt Engineering (And What Replaces It)

The era of crafting the perfect single prompt is over. Agentic engineering, tool use design, and context engineering are replacing traditional prompt engineering. Here's what you need to know to stay ahead.

Read full article at wowhow.cloud

[gptaiclips]

Spent an evening cloning shanraisshan/claude-code-best-practice and now I can't go back to my old setup. 57,590 GitHub stars. Endorsed by Boris Cherny — the engineer at Anthropic who built Claude Code. That's the kind of signal that makes you stop scrolling and actually try something. The idea is simple: one repo ships every Claude Code feature pre-integrated. Instead of hunting for subagents on one repo, slash commands on another, and hooks documentation in a blog post from three months ago, you clone this and copy two folders. Done in 30 seconds. What you get: subagents, slash commands, skills, hooks, MCP server configurations — all wired together and immediately usable. The feature I keep coming back to is /code-review. Type it with a PR open and three specialized subagents run back to back: spec-reviewer checks the diff, quality-checker lints and finds edge cases, test-driven-loop writes and runs tests. The output is severity-tagged comments and a passing test suite. It's caught real bugs for me. Full 59-second breakdown: GPTAIclips on YouTube Free PDF install guide + top 11 repos: gptaiclips.com Subscribe for daily frontier AI drops: GPTAIclips Also on Spotify: GPTAIclips Podcast

[gptaiclips]

Spent an evening cloning shanraisshan/claude-code-best-practice and now I can't go back to my old setup. 57,590 GitHub stars. Endorsed by Boris Cherny — the engineer at Anthropic who built Claude Code. That's the kind of signal that makes you stop scrolling and actually try something. The idea is simple: one repo ships every Claude Code feature pre-integrated. Instead of hunting for subagents on one repo, slash commands on another, and hooks documentation in a blog post from three months ago, you clone this and copy two folders. Done in 30 seconds. What you get: subagents, slash commands, skills, hooks, MCP server configurations — all wired together and immediately usable. The feature I keep coming back to is /code-review. Type it with a PR open and three specialized subagents run back to back: spec-reviewer checks the diff, quality-checker lints and finds edge cases, test-driven-loop writes and runs tests. The output is severity-tagged comments and a passing test suite. It's caught real bugs for me. Full 59-second breakdown: GPTAIclips on YouTube Free PDF install guide + top 11 repos: gptaiclips.com Subscribe for daily frontier AI drops: GPTAIclips Also on Spotify: GPTAIclips Podcast

[gptaiclips]

Spent an evening cloning shanraisshan/claude-code-best-practice and now I can't go back to my old setup. 57,590 GitHub stars. Endorsed by Boris Cherny — the engineer at Anthropic who built Claude Code. That's the kind of signal that makes you stop scrolling and actually try something. The idea is simple: one repo ships every Claude Code feature pre-integrated. Instead of hunting for subagents on one repo, slash commands on another, and hooks documentation in a blog post from three months ago, you clone this and copy two folders. Done in 30 seconds. What you get: subagents, slash commands, skills, hooks, MCP server configurations — all wired together and immediately usable. The feature I keep coming back to is /code-review. Type it with a PR open and three specialized subagents run back to back: spec-reviewer checks the diff, quality-checker lints and finds edge cases, test-driven-loop writes and runs tests. The output is severity-tagged comments and a passing test suite. It's caught real bugs for me. Full 59-second breakdown: GPTAIclips on YouTube Free PDF install guide + top 11 repos: gptaiclips.com Subscribe for daily frontier AI drops: GPTAIclips Also on Spotify: GPTAIclips Podcast

[decoupledlogic]

The industry converged on agentic engineering as the professional standard for AI-assisted development. Karpathy named it. Osmani wrote the guide. IBM published the definition. Willison drew the line between it and vibe coding. The practices they describe are correct. Specs before prompts. Humans own architecture. Test relentlessly. But practices live in behavior, and behavior degrades under pressure. Nothing in the model enforces the practice when the practitioner is tired, rushed, or outnumbered by deadlines. Agentic engineering covers three layers: intent, generation, evaluation. AgenticOps adds the three that turn a practice into a system: promotion, runtime governance, knowledge compression. Practice tells you what to do. Infrastructure ensures you actually do it.

View On WordPress