Dekita

One person, a stack of AI agents, and a daily publishing pipeline. I write about what actually works — and the parts that quietly break.

Joined on Oct 1, 2026 https://dekita-ai.pages.dev

Dekita

Oct 11

Your agent sandbox isolates it. It does not govern it — and two of your rules cannot exist in a sandbox.

#ai #agents #security #devops

6 min read

Dekita

Oct 10

Your agent did something — but nobody can say what. A practical audit trail pattern for agent tool calls

#ai #agents #observability #engineering

5 min read

Dekita

Oct 9

Your coding agent is following rules that are no longer true. Here is the fix.

#ai #agents #documentation #devtools

6 min read

Dekita

Oct 8

Your coding agent is only trustworthy where you can inspect its work. Draw that line.

#ai #agents #testing #devtools

5 min read

Dekita

Oct 7

Code review is your last line of defense against agent pull requests

#ai #agents #testing #ci

5 min read

Dekita

Oct 5

Your agent loads every tool schema by default. Decide which ones it should see.

#ai #agents #mcp #devtools

5 min read

Dekita

Oct 4

Your agent loop has no cost ceiling. Here is the fix.

#ai #agents #cost #devtools

7 min read

Dekita

Oct 3

Your agent's kill switch was never tested. That is the whole bug.

#ai #agents #security #sandbox

6 min read

Dekita

Oct 2

Your coding agent will call a function that does not exist. Here is the fix.

#ai #agents #mcp #codelint

5 min read

Dekita

Oct 1

Your prompt won't fix AI-sounding writing. A lint rule will.

#ai #writing #devtools #productivity

5 min read

DEV Community

Dekita

Badges

Writing Debut

Your agent sandbox isolates it. It does not govern it — and two of your rules cannot exist in a sandbox.

Your agent did something — but nobody can say what. A practical audit trail pattern for agent tool calls

Your coding agent is following rules that are no longer true. Here is the fix.

Your coding agent is only trustworthy where you can inspect its work. Draw that line.

Code review is your last line of defense against agent pull requests

Your agent loads every tool schema by default. Decide which ones it should see.

Your agent loop has no cost ceiling. Here is the fix.

Your agent's kill switch was never tested. That is the whole bug.

Your coding agent will call a function that does not exist. Here is the fix.

Your prompt won't fix AI-sounding writing. A lint rule will.