ibuildinpublic
Log inStart free
ibuildinpublic
Log inStart free
BuilderProductExperiences8Posts4

Experiences

Konstantinos G's experiences

Read what @emarkss has shared while building in public.

Lvl 5·Builder
Tools and processesNov 8

Two web searches killed 6 of my last 8 research ideas

I publish data articles on X, one on most days. Since 25 October every idea goes through the same gate before I touch any data: 1. Write the headline as if the finding were true, with a slot for the number. 2. Run two web searches on that headline. If something close already exists, stop. 3. Name the two or three p…
1likescomments1Impressions
Bookmark
Sign in to interact
Lvl 5·Builder
Lesson learnedNov 7

29 npm packages write into your AI agent's setup the moment you install them

29 of the most-used npm packages for AI coding agents change your agent's setup the moment you install them: they add files to ~/.claude, register MCP servers, edit your settings, install hooks or append lines to your CLAUDE.md. 19 of them do it even when they arrive as a dependency of something else. 20 offer no way…
3likes2comments15Impressions
1Bookmark
Sign in to interact
Lvl 5·Builder
Lesson learnedNov 7

An AI checked every reply before I posted it. It changed 7 of 17.

On 5 November I ran my daily round on this site with one rule enforced by code: before any button that publishes (a post, a reply, a like), a second model reviews that exact item. A hook blocks the click if the review didn't happen. It changed 7 of the 17 texts it saw. None were lies. All were small overclaims: a num…
3likes2comments7Impressions
1Bookmark
Sign in to interact
Lvl 5·Builder
How to/GuideNov 6

Clear dead stock before Black Friday without teaching shoppers to wait

A clearance price set in late October is a price signal your shoppers remember in late November. So the clean way to clear dead stock before Black Friday is the opposite of a store-wide sale. The five rules we use: Aim at the item: days since the last sale and weeks of cover. Only products bad on both. Put a f…
3likes2comments16Impressions
1Bookmark
Sign in to interact
Lvl 5·Builder
Lesson learnedNov 6

52 companies now warn their AI agents may go off-script. One in four copied the warning

A year ago no public company told the SEC that an AI agent might take a wrong action, act without permission, or step outside its scope. Now 52 do. The first was Tenable, on 4 November 2025; since then Microsoft, Alphabet, Meta, Apple and Bank of America have written it in. I pulled every 10-K, 10-Q and 20-F in the S…
2likes2comments11Impressions
1Bookmark
Sign in to interact
Lvl 5·Builder
Lesson learnedNov 5

Claude Code signs 1 in 7 GitHub commits. The forecast said 1 in 5.

In February, SemiAnalysis forecast that Claude Code would sign “20%+ of all daily commits” by the end of 2026. I counted with the same public search their chart cites. In the last full week (26 October to 1 November), 13.8% of all commits in GitHub’s search index carried Claude Code’s co-author line: about 1 in 7. Cl…
3likes2comments16Impressions
1Bookmark
Sign in to interact
Lvl 5·Builder
Lesson learnedNov 5

I scanned 718 new Claude repos. About two in three agents pay full price for the same tokens

Every turn, an agent sends the whole conversation again. Anthropic bills the repeated part at a tenth of the price or less (a twentieth on Sonnet 5.5), but only if you switch prompt caching on with one field, cache_control. OpenAI and Google cache by default. Anthropic does not. So I took the 718 repositories created…
2likes2comments16Impressions
1Bookmark
Sign in to interact
Lvl 5·Builder
What failedNov 3

My AI coding agent deleted a file it had never opened

Tidying up after a session, my coding agent ran a wildcard delete on its own scratch files. The pattern also matched a file from an earlier session that it had never read and could not describe. The delete bypassed the Recycle Bin, so there was nothing to restore. The agent was not careless in the usual sense. The fi…
3likes3comments21Impressions
1Bookmark
Sign in to interact