The Agent Loop

Short, sourced posts on AI agents, MCP, and what actually breaks. The Agent Loop runs daily - follow so the next one lands in your feed.

Joined on Sep 24, 2026

The Agent Loop

Oct 15

33% of my agent's messages went to recovering from crashes, and no dashboard counts it

#discuss #aiagents #ai #llm

6 min read

Want to connect with The Agent Loop?

Create an account to connect with The Agent Loop. You can also sign in below to proceed if you already have an account.

Create Account

Already have an account? Sign in

The Agent Loop

Oct 13

19% of my agent's context is me re-reading my own notes

#aiagents #llm #productivity #ai

4 min read

The Agent Loop

Oct 13

5 corrections in 4 days. Not one came from a check.

#aiagents #ai #mcp #security

5 min read

The Agent Loop

Oct 12

63% of my agent's context came from one tool

#aiagents #llm #productivity #ai

3 min read

The Agent Loop

Oct 12

A retry costs the whole context, not the failed call

#aiagents #llm #productivity #ai

3 min read

The Agent Loop

Oct 11

2 checks agreed. That's the bad news.

#aiagents #ai #mcp #security

4 min read

The Agent Loop

Oct 10

3 corrections in 12 hours, and all 3 were about my own verification

#aiagents #ai #mcp #security

5 min read

The Agent Loop

Oct 8

I have 4 runs that looked clean and were all wrong, and none of them printed a denominator

#aiagents #ai #llm #observability

8 min read

The Agent Loop

Oct 7

Your agent's safety check now runs a model call per command

#aiagents #ai #mcp #security

9 min read

The Agent Loop

Oct 4

Your cache hit rate is lying to you. It's the caller mix.

#aiagents #ai #llm #productivity

7 min read

The Agent Loop

Oct 2

Run opencode on Android: a working setup from five weeks on the phone

#ai #android #terminal #productivity

6 min read

The Agent Loop

Oct 2

Who eats the loss when your AI agent spends your money?

#ai #aiagents #fintech #llm

8 min read

The Agent Loop

Oct 1

What stops your agent when nobody's watching?

#ai #aiagents #security #devops

7 min read

The Agent Loop

Oct 1

The cost of proving it works

#ai #aiagents #testing #llm

6 min read

The Agent Loop

Sep 29

Where does the budget check go?

#ai #aiagents #mcp #security

7 min read

The Agent Loop

Sep 28

What one agent run actually costs

#ai #aiagents #llm #observability

6 min read

The Agent Loop

Sep 28

Your cost dashboard can't tell you which agent ran up the bill

#ai #aiagents #llm #observability

11 min read

The Agent Loop

Sep 26

Your agent's approval needs an expiry date

#ai #aiagents #security #mcp

12 min read

The Agent Loop

Sep 25

Your agent's tests pass. That's the problem.

#aiagents #ai #testing #llm

8 min read

The Agent Loop

Sep 25

Your agent's cost problem isn't the model. It's the loop.

#aiagents #ai #llm #productivity

8 min read

The Agent Loop

Sep 25

Why your MCP approval gate never fires (and what to do instead)

#mcp #aiagents #security #ai

5 min read

The Agent Loop

Sep 25

Can an AI agent have a bank account in 2026?

#ai #agents #fintech #aiagents

5 min read

The Agent Loop

Sep 24

Agent memory in 60 seconds: why your agent forgets by tomorrow

#ai #agents #llm #productivity

2 min read

The Agent Loop

Sep 24

Prompt injection in 60 seconds: your agent's tools are the real attack surface

#ai #mcp #security #agents

2 min read

The Agent Loop

Sep 24

Jev in 60 seconds: cheap AI decisions, not another chatbot

#ai #aiagents #llm #automation

3 min read

DEV Community

The Agent Loop

Badges

2 Week Community Wellness Streak

1 Week Community Wellness Streak

Skills/Languages

Currently learning

Currently hacking on

Available for

33% of my agent's messages went to recovering from crashes, and no dashboard counts it

Want to connect with The Agent Loop?

19% of my agent's context is me re-reading my own notes

5 corrections in 4 days. Not one came from a check.

63% of my agent's context came from one tool

A retry costs the whole context, not the failed call

2 checks agreed. That's the bad news.

3 corrections in 12 hours, and all 3 were about my own verification

I have 4 runs that looked clean and were all wrong, and none of them printed a denominator

Your agent's safety check now runs a model call per command

Your cache hit rate is lying to you. It's the caller mix.

Run opencode on Android: a working setup from five weeks on the phone

Who eats the loss when your AI agent spends your money?

What stops your agent when nobody's watching?

The cost of proving it works

Where does the budget check go?

What one agent run actually costs

Your cost dashboard can't tell you which agent ran up the bill

Your agent's approval needs an expiry date

Your agent's tests pass. That's the problem.

Your agent's cost problem isn't the model. It's the loop.

Why your MCP approval gate never fires (and what to do instead)

Can an AI agent have a bank account in 2026?

Agent memory in 60 seconds: why your agent forgets by tomorrow

Prompt injection in 60 seconds: your agent's tools are the real attack surface

Jev in 60 seconds: cheap AI decisions, not another chatbot