docs: add new entries and updates to AI coding history and concepts

codekiln · codekiln · commit 2202845814a1 · 2025-10-19T07:31:06.000-04:00
- Introduced new journal entry for "Agentic AI Coding" with updates on related topics.
- Added detailed pages for phases 5 of AI coding, including "Context Engineering" and its sub-phases.
- Created new entries for "Vibe Coding" and clarified its distinction from traditional coding practices.
- Updated references and embedded links for better navigation and context within the AI coding documentation.
diff --git a/journals/2025_10_19.md b/journals/2025_10_19.md
@@ -0,0 +1,8 @@
+- [[Agentic AI Coding]]
+	- #Updated
+		- [[AI/Coding/History]] and subpages
+	- #Filed
+		- [[Person/Simon Willison/Blog/25/10/Claude Skills]]
+		- [[LobsteRS]]
+		- [[Person/Peter Steinberger/Blog/25/10/Just Talk to It - no-bs Agentic Engineering]]
+		- [[Person/Peter Steinberger/GitHub/Gist/Agent.md]]
diff --git a/pages/AI___Coding___Agentic.md b/pages/AI___Coding___Agentic.md
@@ -2,11 +2,5 @@ alias:: [[Agentic AI Coding]]
 tags:: [[Term]]
 
 - # Agentic AI Coding
-	- AI coding has evolved in phases
-		- Phase 1 -
-		- Phase 2 - using [[ChatGPT]] with [[Copy/and Paste]] - a step beyond,
-		- Phase 3 - single-file editing in [[IDE]]
-		- Phase 4 - multi-file editing in [[IDE]]
-		- Phase 5 - [[Vibe Coding]] becomes possible with agentic multi-file editing both in [[IDE]] and with tools like [[BoltAI]] and [[LoveableAI]]
-		- Phase 6 - bringing remote context into coding agents with [[MCP]]
-		- Phase 7 - dynamic context engineering with [[Claude Code Skills]]
+	- An LLM running in a loop with tools to create code.
+	- See [[AI/Coding/History/Phase/4 - Basic Autonomous Coding]] and subsequent phases like [[AI/Coding/History/Phase/5 - Context Engineering]]
diff --git a/pages/AI___Coding___History.md b/pages/AI___Coding___History.md
@@ -3,5 +3,4 @@
 -
 - {{embed [[AI/Coding/History/Phase/3 - Direct Agentic Coding]]}}
 - {{embed [[AI/Coding/History/Phase/4 - Basic Autonomous Coding]]}}
-- [[AI/Coding/History/Phase/5 - Context Engineering/MCP Servers]]
-- [[AI/Coding/History/Phase/5 - Context Engineering/Skill Bundles]]
+- {{embed [[AI/Coding/History/Phase/5 - Context Engineering]]}}
diff --git a/pages/AI___Coding___History___Phase___4 - Basic Autonomous Coding.md b/pages/AI___Coding___History___Phase___4 - Basic Autonomous Coding.md
@@ -1,5 +1,6 @@
 - Ushered in by [[Claude Code]], in phase four, a developer could give an [[LLM]] an instruction, and an [[Agentic System]] would iterate on it for some time, sometimes 3-5 minutes, other times 5-10 minutes.
-- [[Person/Andrej Karpathy]] coined the term [[Vibe Coding]]
+- {{embed [[AI/Coding/Vibe/25/02/Karpathy Coins Term]]}}
+- {{embed [[AI/Coding/Vibe/25/03/Willison Clarifies as Not Reviewing Code]]}}
 - Sub-Phases
 	- {{embed [[AI/Coding/History/Phase/4 - Basic Autonomous Coding/Web-based Vibe Coding]]}}
 	- {{embed [[AI/Coding/History/Phase/4 - Basic Autonomous Coding/CLI-based]]}}
diff --git a/pages/AI___Coding___History___Phase___5 - Context Engineering.md b/pages/AI___Coding___History___Phase___5 - Context Engineering.md
@@ -0,0 +1,4 @@
+- After agentic coding became the dominant paradigm, it took a while for engineers to understand how to let the agentic coding tools take actions in outside of the local filesystem, such as checking CI logs, reading JIRA or GitHub tickets, or opening PRs. After the introduction of [[MCP]] made this possible, there was a period of time where  developers struggled to understand how agentic tools quickly filled up the [[AI/Context/Window]] and led to [[Context Rot]]. The term [[Context Engineering]] emerged as a way to signal the importance for minimizing the context needed to solve the problem at hand while still providing access to powerful tools.
+- Sub-Phases
+	- {{embed [[AI/Coding/History/Phase/5 - Context Engineering/MCP Servers]]}}
+	- {{embed [[AI/Coding/History/Phase/5 - Context Engineering/Skill Bundles]]}}
diff --git a/pages/AI___Coding___History___Phase___5 - Context Engineering___MCP Servers.md b/pages/AI___Coding___History___Phase___5 - Context Engineering___MCP Servers.md
@@ -0,0 +1,4 @@
+- After March 2025, many [[IDE]]s such as [[CursorAI]] and some [[CLI]]s such as [[Claude Code]] implemented an [[MCP/Client]] into the agentic coding chat interface, so developers could give AI [[MCP Servers]] to which enabled them to ask the AI to both sense and act on external context, for example
+	- [[GitHub/MCP]] included the ability open and update [[PR]]s, to read [[CICD]] logs and [[GitHub/Issue]]s
+	- [[Atlassian/MCP]] included the ability to edit and read [[Confluence]], [[JIRA]]
+	- [[Playwright/MCP]] included the ability to check out a web application's display looked like in the browser, giving it the ability to "see"
diff --git a/pages/AI___Coding___History___Phase___5 - Context Engineering___Skill Bundles.md b/pages/AI___Coding___History___Phase___5 - Context Engineering___Skill Bundles.md
@@ -0,0 +1 @@
+- [[Person/Simon Willison/Blog/25/10/Claude Skills]]
diff --git a/pages/AI___Coding___Vibe.md b/pages/AI___Coding___Vibe.md
@@ -3,13 +3,9 @@ tags:: [[Term]]
 
 - # Vibe Coding
 	- ## #History of Vibe Coding as a Term and an Idea
-		- On [[2025-10-02 Thu]], [[Person/Andrej Karpathy]] coined the term "Vibe Coding" in a [tweet](https://x.com/karpathy/status/1886192184808149383).
-			- > There’s a new kind of coding I call “vibe coding”, where you fully give in to the vibes, embrace exponentials, and **forget that the code even exists**. It’s possible because the LLMs (e.g. Cursor Composer w Sonnet) are getting too good. Also I just talk to Composer with SuperWhisper so I barely even touch the keyboard.
-			- > I ask for the dumbest things like “decrease the padding on the sidebar by half” because I’m too lazy to find it. I “Accept All” always, I don’t read the diffs anymore. When I get error messages I just copy paste them in with no comment, usually that fixes it. The code grows beyond my usual comprehension, I’d have to really read through it for a while. Sometimes the LLMs can’t fix a bug so I just work around it or ask for random changes until it goes away.
-			- > **It’s not too bad for throwaway weekend projects, but still quite amusing**. I’m building a project or webapp, but it’s not really coding—I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works.
+		- {{embed [[AI/Coding/Vibe/25/02/Karpathy Coins Term]]}}
 		- On [[2025-02-25 Tue]], I had my first real feeling of this [[CursorAI]], downstream from [[Person/Harper Reed/Blog/25/02/My LLM codegen workflow atm]].
-		- On [[2025-03-19 Wed]], [[Person/Simon Willison/Blog/25/03/Not all AI-assisted programming is vibe coding]], clarified
-			- > When I talk about vibe coding I mean **building software with an LLM without reviewing the code it writes**
+		- {{embed [[AI/Coding/Vibe/25/03/Willison Clarifies as Not Reviewing Code]]}}
 		- On [[2025-10-08 Wed]], [[Person/Simon Willison/Blog/25/10/Vibe Engineering]] suggested that [[Vibe Coding]] is the opposite of a new term [[Vibe Engineering]]:
 			- > the other end of the spectrum from [[Vibe Coding]], where **seasoned professionals accelerate their work with LLMs while staying proudly and confidently accountable for the software they produce**
 	- Related Concepts:
diff --git a/pages/AI___Coding___Vibe___25___02___Karpathy Coins Term.md b/pages/AI___Coding___Vibe___25___02___Karpathy Coins Term.md
@@ -0,0 +1,4 @@
+- On [[2025-10-02 Thu]], [[Person/Andrej Karpathy]] coined the term "Vibe Coding" in a [tweet](https://x.com/karpathy/status/1886192184808149383).
+	- > There’s a new kind of coding I call “vibe coding”, where you fully give in to the vibes, embrace exponentials, and **forget that the code even exists**. It’s possible because the LLMs (e.g. Cursor Composer w Sonnet) are getting too good. Also I just talk to Composer with SuperWhisper so I barely even touch the keyboard.
+	- > I ask for the dumbest things like “decrease the padding on the sidebar by half” because I’m too lazy to find it. I “Accept All” always, I don’t read the diffs anymore. When I get error messages I just copy paste them in with no comment, usually that fixes it. The code grows beyond my usual comprehension, I’d have to really read through it for a while. Sometimes the LLMs can’t fix a bug so I just work around it or ask for random changes until it goes away.
+	- > **It’s not too bad for throwaway weekend projects, but still quite amusing**. I’m building a project or webapp, but it’s not really coding—I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works.
diff --git a/pages/AI___Coding___Vibe___25___03___Willison Clarifies as Not Reviewing Code.md b/pages/AI___Coding___Vibe___25___03___Willison Clarifies as Not Reviewing Code.md
@@ -0,0 +1,2 @@
+- On [[2025-03-19 Wed]], [[Person/Simon Willison/Blog/25/03/Not all AI-assisted programming is vibe coding]], clarified
+	- > When I talk about vibe coding I mean **building software with an LLM without reviewing the code it writes**
diff --git a/pages/AI___Context___Window.md b/pages/AI___Context___Window.md
@@ -0,0 +1,5 @@
+tags:: [[Term]]
+alias:: [[AI Context Window]], [[AI/LLM/Context/Window]]
+
+- AI Context Window
+	- [[LLMs]] have a "context window" that is determined by a maximum number of input [[Token]]s. For a given model, this window affects the upper bound of the complexity of the problem they can handle as well as the amount of private business data that can be utilized in a query.
diff --git a/pages/BetterStack.md b/pages/BetterStack.md
@@ -0,0 +1 @@
+- [Better Stack - Observability Meets Incident Management](https://betterstack.com/)
diff --git a/pages/LobsteRS.md b/pages/LobsteRS.md
@@ -0,0 +1,2 @@
+- https://lobste.rs/
+	- This is a community that's been around since [[2013]] that reminds me of [[HackerNews]]. Their [about page](https://lobste.rs/about) signals the intentionality of this community.
diff --git a/pages/Person___Jesse Vincent___Blog___25___10___Superpowers Oct 2025.md b/pages/Person___Jesse Vincent___Blog___25___10___Superpowers Oct 2025.md
@@ -0,0 +1,3 @@
+# [Superpowers: How I'm using coding agents in October 2025](https://blog.fsck.com/2025/10/09/superpowers/)
+	- Documents how Jesse was working on a new feature "Superpowers" for his own AI coding workflow, and then [[Claude Code/Skill]]s came out, making this basically
+		- > I've spent the past couple of weeks working on a set of tools to better extract and systematize my processes and to help better steer my agentic buddy. I'd been planning to start to document the system this weekend, but then this morning, Anthropic went and rolled out [a plugin system for claude code](https://docs.claude.com/en/docs/claude-code/plugins).
diff --git a/pages/Person___Peter Steinberger___Blog___25___10___Just Talk to It - no-bs Agentic Engineering.md b/pages/Person___Peter Steinberger___Blog___25___10___Just Talk to It - no-bs Agentic Engineering.md
@@ -0,0 +1,42 @@
+# [Just Talk To It - the no-bs Way of Agentic Engineering | Peter Steinberger](https://steipete.me/posts/just-talk-to-it?__readwiseLocation=)
+	- Header graphic credit https://x.com/thorstenball/status/1976224756669309195
+		- ![The Curve of Agentic Programming - x axis is Time and y axis is Complexity and something like a bell curve reveals that the author started with "plz fix this" then went through a more complex mode of "8 agents at the same time with complex orchestration" and then finally went to just asking the AI to do things like "Hey, look at these files and then do [med size change]"](https://steipete.me/assets/img/2025/just-talk-to-it/curve-angentic.jpg)
+	- Fantastic overview of the [[SOTA]] for [[Agentic AI Coding]]
+	- He has abandoned [[Claude Code]] for [[OpenAI/Codex/CLI]] because
+		- Codex CLI has `~230k` usable context vs `~156k` for Claude Code, and token use is more efficient
+		- Codex CLI is written in rust and it's faster
+		- the language Claude Code use is bad for his "mental health" - the "Absolutely right's," the "100% production ready," the gaslighting, etc, and he finds that Codex CLI has none of this.
+		- #Wow ==Peter is paying $1k per month for "4 OpenAI subs and 1 Anthropic sub"== in order to have "unlimited tokens" and he's got a lot of skin in the game. He says,
+			- > IMO there's simply not much space between the end user and the model company
+			- I'm not totally sure what he means, but this was said in a paragraph where he was making the point that if he coded with API calls it would cost "around 10x more," so basically subscriptions are the only way to go.
+	- He gives a nice overview of many tools,
+		- [[CursorAI]] - still has bugs from back in May
+		- [[Auggie]] - blip on timeline, nobody mentioned again
+		- > [[RAG]] might have been helpful for [[Anthropic/Model/Claude/Sonnet]] but [[gpt-5]] is so good at searching that you don't need a separate vector index for your code
+			- [[My Note]] - *why is it that people are not talking about vector indexes for documentation?*
+		- [[Open/Weights]] models like [[GLM/4.6]] and [[Kimi/K2.1]] "are strong contendors that slowly reach [[Anthropic/Model/Claude/Sonnet/3.7]] quality
+		- Web tools like [[Devin]], [[OpenAI/Codex/Web]], [[JulesAI]] from [[Google]],
+		- [[ConductorAI]], [[Claude Code Terragon]], [[SculptorAI]], "and the 1000 other ones. Some are hobby projects, some are drowning in VC money ... None stick."
+	- He uses [[Mac/App/WisprFlow]] for [[Speech-to-Text]]
+		- > Wispr Flow with semantic correction is still king.
+	- [[AI/Coding/Background Task]]s
+		- He reverted back to [[Claude Code]] for a while because of its support for background tasks, which [[OpenAI/Codex/CLI]] doesn't have yet, but now he just uses [[tmux]] instead and it can run CLIs in persistent sessions in the background, and there's plenty of knowledge about how to do that in the model, so you just need to say "run via tmux" great #Tip !
+	- [[MCP Server]]s
+		- > IMO most are something for the marketing department to make a checkbox and be proud. ==Almost all MCPs really should be clis==. I say that as someone who wrote 5 MCPs myself.
+			- #Example - [steipete/claude-code-mcp: Claude Code as one-shot MCP server to have an agent in your agent.](https://github.com/steipete/claude-code-mcp) #LOL
+		- > I can just refer to a [[CLI/Tool]] by name. I don’t need any explanation in my [[Agents.md]]. The agent will try `$randomcrap` on the first call, the cli will present the help menu, context now has full info how this works and from now on we good. I don’t have to pay a price for any tools, unlike [[MCP]]s which are a **constant cost and garbage in my context**. Use [[GitHub/MCP]] and see **23k tokens gone**. Heck, they did make it better because it was almost 50.000 tokens when it first launched. Or use the [[gh]] cli **which has basically the same feature set, models already know how to use it, and pay zero context tax**.
+		- He introduced me to some [[Observability]] tools that were [[New/To/Me]]
+			- [[Inngest]] [AI and backend workflows, orchestrated at any scale](https://www.inngest.com/) - a bit hard to tell, but if I squint, it looks like a [[LangSmith]] and [[langgraph/Platform]] competitor.
+			- [[BetterStack]] [Better Stack - Observability Meets Incident Management](https://betterstack.com/) Looks like a DataDog / Sentry / Splunk / New Relic competitor
+		- He says he uses [[Chrome/DevTools/MCP]] to replace [[Playwright/MCP]] "as my go-to MCP for web debugging ... I don't need to do it lots but when I do, it's quite useful to close the loop."
+			- > I **designed my website** so that I can **create api keys** that **allow my model to query any endpoint via curl**, which is faster and more token-efficient in almost all use cases, so even that MCP isn’t something I need daily.
+				- *I've not heard about this [[AI/Coding/Technique]] before... I'd love to read a more detailed description*
+	- #Refactoring
+		- [[New/To/Me]]
+			- [[jscpd]]
+			- [[knipDev]]
+		- > mentioning modern react patterns and rewriting code, e.g. [You Might Not Need an Effect – React](https://react.dev/learn/you-might-not-need-an-effect) #React
+			- also, uses [[ESLint]]'s [[React/Compiler]]
+	- ## [[AI/Coding/Technique/Spec-Driven Dev]]
+		- in [[2025/06]] he was using this [The Future of Vibe Coding: Building with AI, Live and Unfiltered | Peter Steinberger](https://steipete.me/posts/2025/the-future-of-vibe-coding)
+		- Now he says that the technique is not worth the time, and starts in chatgpt and moves to code later.
diff --git a/pages/Person___Peter Steinberger___GitHub___Gist___Agent.md.md b/pages/Person___Peter Steinberger___GitHub___Gist___Agent.md.md
@@ -0,0 +1 @@
+- [Agent rules for git](https://gist.github.com/steipete/d3b9db3fa8eb1d1a692b7656217d8655)
diff --git a/pages/Person___Simon Willison___Blog___25___10___Claude Skills.md b/pages/Person___Simon Willison___Blog___25___10___Claude Skills.md
@@ -0,0 +1,4 @@
+tags:: [[Claude Code Skills]]
+
+- # [simonw/claude-skills](https://simonwillison.net/2025/Oct/10/claude-skills/)
+	- Quotes [[Person/Jesse Vincent/Blog/25/10/Superpowers Oct 2025]]
diff --git a/pages/Person___Tom Doerr.md b/pages/Person___Tom Doerr.md
@@ -1,4 +1,4 @@
-tags:: [[Person]], [[AI Engineer]], [[GitHub]], [[OpenAI]], [[Codex]], [[ZSH]], [[Vim]], [[Python]]
+tags:: [[Person]], [[AI Engineer]]
 
 - # Tom Dörr
 	- ## Overview
diff --git a/pages/SOTA.md b/pages/SOTA.md
@@ -0,0 +1,4 @@
+tags:: [[Term]], [[Abbreviation]]
+
+- # State of the Art (SOTA)
+	-
diff --git a/pages/jscpd.md b/pages/jscpd.md
@@ -0,0 +1,4 @@
+tags:: [[Refactoring]], [[CLI/Tool]]
+
+- [kucherenko/jscpd: Copy/paste detector for programming source code.](https://github.com/kucherenko/jscpd)
+	-
diff --git a/pages/knipDev.md b/pages/knipDev.md
@@ -0,0 +1,7 @@
+tags:: [[CLI/Tool]]
+
+- [Declutter your JavaScript & TypeScript projects | Knip](https://knip.dev/)
+	- > Knip finds and fixes unused dependencies, exports and files. Use it for enhanced code and dependency management.
+	- > Advanced analysis starting from fine-grained entry points based on the actual frameworks and tooling in (mono)repos for accurate and actionable results.
+	- > Knip comes with [100+ plugins](https://knip.dev/reference/plugins) for tools and frameworks like Astro, Cypress, ESLint, Jest, GitHub Actions, Next.js, Nx, Remix, Storybook, Svelte, Vite, Vitest, Webpack and many, many more.
+	-

Original file line number	Diff line number	Diff line change
`@@ -3,5 +3,4 @@`
`3`	`3`	`-`
`4`	`4`	`- {{embed [[AI/Coding/History/Phase/3 - Direct Agentic Coding]]}}`
`5`	`5`	`- {{embed [[AI/Coding/History/Phase/4 - Basic Autonomous Coding]]}}`
`6`		`-- [[AI/Coding/History/Phase/5 - Context Engineering/MCP Servers]]`
`7`		`-- [[AI/Coding/History/Phase/5 - Context Engineering/Skill Bundles]]`
	`6`	`+- {{embed [[AI/Coding/History/Phase/5 - Context Engineering]]}}`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1 @@`
	`1`	`+- [[Person/Simon Willison/Blog/25/10/Claude Skills]]`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,2 @@`
	`1`	`+- On [[2025-03-19 Wed]], [[Person/Simon Willison/Blog/25/03/Not all AI-assisted programming is vibe coding]], clarified`
	`2`	`+ - > When I talk about vibe coding I mean building software with an LLM without reviewing the code it writes`