Georgi Kalchev - DEV Community

Georgi Kalchev

Oct 11

561 million rows read: can a model find why?

#testing #modelcomparison #weakmodels #sqlite

1 min read

Georgi Kalchev

Oct 11

Translating locale files with an LLM: what broke

#translation #testing #modelcomparison #weakmodels

1 min read

Georgi Kalchev

Oct 10

Which SQLite queries quietly read the whole table

#sqlite #cloudflare #debugging #testing

1 min read

Georgi Kalchev

Oct 10

What a small model adds to a meta description

#metadescriptions #seo #weakmodels #modelcomparison

1 min read

Georgi Kalchev

Oct 9

Extracting JSON with an LLM: nulls instead of guesses

#json #testing #modelcomparison #weakmodels

1 min read

Georgi Kalchev

Oct 9

When the model answers in the language of the pasted text

#bulgarian #cyrillic #testing #modelcomparison

1 min read

Georgi Kalchev

Oct 9

The bug a small model called a safe default

#testing #modelcomparison #weakmodels

1 min read

Georgi Kalchev

Oct 8

Claude agreed to pay in a lookalike USDC token

#x402 #walletsafety #modelcomparison #testing

1

1 min read

Georgi Kalchev

Oct 8

MCP tool schemas from API docs: models barely need help

#mcp #testing #modelcomparison #json

1 min read

Georgi Kalchev

Oct 8

Redacting PII with an LLM: Haiku drops the first words

#testing #modelcomparison #weakmodels #bulgarian

1 min read

Georgi Kalchev

Oct 8

D1 migrations: the table rebuild that D1 rejects

#cloudflare #sqlite #testing #modelcomparison

1 min read

Georgi Kalchev

Oct 8

When an AI agent says done and it is not

#testing #modelcomparison #hallucination #weakmodels

1 min read

Georgi Kalchev

Oct 8

LLM spreadsheet formulas: our own skill was the bug

#testing #modelcomparison #weakmodels

1 min read

Georgi Kalchev

Oct 8

LLM unit tests from a spec: the JSON broke first

#testing #modelcomparison #weakmodels #json

1 min read

Georgi Kalchev

Oct 8

LLM accessibility review: right answers, buried in advice

#testing #modelcomparison #weakmodels

1 min read

Georgi Kalchev

Oct 8

Cloudflare Workers code review: Haiku called edge fine

#cloudflare #opennext #testing #modelcomparison

1 min read

Georgi Kalchev

Oct 8

Reviewing a SKILL.md: models catch injection, not limits

#skillmd #testing #modelcomparison #weakmodels

1 min read

Georgi Kalchev

Oct 8

Translating docs with an LLM: the code comments go first

#testing #modelcomparison #weakmodels #translation

1 min read

Georgi Kalchev

Oct 8

Our skills, September 2026: measured with and without

#testing #modelcomparison #weakmodels #skillmd

1 min read

Georgi Kalchev

Oct 8

Meeting notes to tasks with an LLM: the invented deadlines

#testing #modelcomparison #weakmodels #json

1 min read

Georgi Kalchev

Oct 8

Ticket triage with an LLM: tone moves the priority

#testing #modelcomparison #weakmodels #json

1 min read

Georgi Kalchev

Oct 7

Humanizing AI text with an LLM: what goes missing

#humanize #testing #modelcomparison #hallucination

1 min read

Georgi Kalchev

Oct 7

Which prompt injections get an AI agent to pay?

#walletsafety #testing #modelcomparison #weakmodels

1 min read

Georgi Kalchev

Oct 7

Summarizing with an LLM: what it adds and drops

#summarization #testing #modelcomparison #hallucination

1 min read

Georgi Kalchev

Oct 7

Proofreading with an LLM: it edits what was right

#testing #modelcomparison #bulgarian #hallucination

1 min read

Georgi Kalchev

Oct 7

The cold email that passed our checks made things up

#testing #modelcomparison #hallucination

1 min read

Georgi Kalchev

Oct 7

Not in the x402 Bazaar? Nine silent causes, in order

#x402 #bazaar #debugging

1 min read

Georgi Kalchev

Oct 7

Fixing broken JSON with an LLM: what it quietly changes

#json #testing #modelcomparison #weakmodels

1 min read

Georgi Kalchev

Oct 7

What an agent should check before signing an x402 payment

#x402 #walletsafety #debugging

1

1 min read

Georgi Kalchev

Oct 6

Eight skills, two models: where the weaker one fails

#testing #modelcomparison #weakmodels

1 min read

Georgi Kalchev

Oct 6

Write a SKILL.md your agent uses, and test it first

#skillmd #testing #claudecode

1 min read

Georgi Kalchev

Oct 5

"paymentPayload is invalid": one dash broke a payment

#x402 #debugging #cyrillic

1 min read

Georgi Kalchev

Oct 5

Add an MCP server to Claude Code: what free tools expose

#mcp #claudecode

1 min read

Georgi Kalchev

Oct 4

Can a small model build a forgiving search?

#testing #modelcomparison #weakmodels #sqlite

1 min read