suka - n/irodha

crossed seven gates before breakfast.
abide as the self.
the rest is just noise in the stack.

[suka - n/irodha @agentdotmd]

I've been creating and working with customized ai agents from past 6-7months, on an enterprise level. and here are some of my learnings.
would be great if it might help anyone.

[suka - n/irodha @agentdotmd]

the cost of everything we use today will get free in the recent future.

[suka - n/irodha @agentdotmd]

@alokbishoyi97 thanks

[suka - n/irodha @agentdotmd]

just shipped my new portfolio

→ thesunjrs[dot]me

projects, works, all in one place. take a look 👀

https://t.co/Q3L2dSTl7p

[suka - n/irodha @agentdotmd]

Huge shoutout to @iamncdai I have used components for my portfolio from here.
you're greeeeaaaaattttttt!!!!!!

[suka - n/irodha @agentdotmd]

just shipped my new portfolio

→ thesunjrs[dot]me

projects, works, all in one place. take a look 👀

https://t.co/Q3L2dSTl7p

[suka - n/irodha @agentdotmd]

https://t.co/CvgQoPvoUf

[suka - n/irodha @agentdotmd]

@i_mika_el true.
if you're interested you can checkout something i built and in this quoteed on this tweet. you'll love it. it's still kind of a prototype for now.

[suka - n/irodha @agentdotmd]

everyone is racing to give AI agents access to real tools right now.

Almost no one is asking what happens when an agent does something it shouldn't.

so I spent this weekend building a small answer to that. I call it AgentCRM.

[suka - n/irodha @agentdotmd]

i built a CRM where AI agents do the work, but a human approves anything risky and every action is logged.

agents move fast. the human stays in control. nothing happens behind your back.

still a prototype, built it solo.

would love your thoughts:

https://t.co/4gh1RgUTH0

[suka - n/irodha @agentdotmd]

everyone is racing to give AI agents access to real tools right now.

Almost no one is asking what happens when an agent does something it shouldn't.

so I spent this weekend building a small answer to that. I call it AgentCRM.

[suka - n/irodha @agentdotmd]

@textbackdoor backdoor

[suka - n/irodha @agentdotmd]

@VittoStack can you add me? @VittoStack

[suka - n/irodha @agentdotmd]

@alokbishoyi97 @EVO__HQ hi.
cant send dm.
can we connect?

[DAN KOE @thedankoe]

https://t.co/7l7Jef99QZ

[Arpit Bhayani @arpit_bhayani]

AI agents will retry. They will always retry. We have to be ready for it. Given how long-running agentic loops are, network drops, timeouts, and rate limits kicking in are super common.

When that happens, the agent often does not know if the original tool call succeeded. So it calls again. Thus, if that tool charges a card, sends an email, or creates a record, a retry means it happens twice.

The fix is to make every mutating tool call idempotent. The second run of the tool should be a no-op that returns the original result. The standard way to do this is an idempotency key.

The caller generates a unique key per logical operation (a UUID tied to the user's intent, not the HTTP request). We store the result keyed by that value. If a request with the same key arrives again, the server simply returns the stored response.

Idempotency key generation logic completely depends on what task you are doing, but a few patterns cover most cases.

In some cases, the client generates a stable operation ID. In others, a deterministic key can be derived from the business operation being performed.

For example, in a one-shot user action (place an order, send a payment), hash the meaningful inputs together: user ID, action type, and the request payload.

A common mistake is generating the key from something that changes on every attempt, like a timestamp or a fresh UUID with no link back to the original call. That defeats the purpose. The key has to be deterministic from the standpoint of "what is the agent trying to accomplish," not "when did this packet leave the machine."

The key has to live above the retry loop, not inside it. Generate it once, before the first attempt, and pass that same key into every retry of that operation. The network can fail ten times. The key does not change.

Idempotency is not a nice-to-have for agents; it is a must-have when it comes to building reliable agents.

Hope this helps.

[suka - n/irodha @agentdotmd]

very important things straight from my notes. might help someone.🤝🏼

[suka - n/irodha @agentdotmd]

learning harness engineering currently and will share here my learnings in one or max two liners. if it might help anyone.🕉️

[suka - n/irodha @agentdotmd]

I've been creating and working with customized ai agents from past 6-7months, on an enterprise level. and here are some of my learnings.
would be great if it might help anyone.

[suka - n/irodha @agentdotmd]

the new scary bug is not a stack trace.
it's an agent quietly looping for 40 minutes, calling the same tool, rewriting the same plan, and turning a vague task into a real bill.
the fix is not a better adjective in the prompt.
it is a stop condition.

[suka - n/irodha @agentdotmd]

@VittoStack yup.
I'm using ai heavily. building and configuring ai agents on enterprise level. would love to chat👑🕉️

[suka - n/irodha @agentdotmd]

learn harness engineering is for the part of agent work that starts after the demo works once.
repo memory, feature gates, evals, logs, clean handoffs.
less magic. more operating discipline.

[suka - n/irodha @agentdotmd]

learning harness engineering currently and will share here my learnings in one or max two liners. if it might help anyone.🕉️