Suraj Kumar

Building swarm-test -- chaos engineering for multi - agent AI systems . AI systems Architect. Building the reliability layer for the agent economy

[Suraj Kumar @SurajKumar__10]

Your multi-agent system has a defect in it right now, and your tests will never find it.

Not a hypothetical. A structural one: a loop with no exit, an agent everything silently depends on, an untrusted tool output reaching a decision agent with no validation.

It's in there today. It throws no error. It fires in production, not in testing — which is why you'll meet it on the invoice, in an incident channel, at 2am.

14 agents at 95% reliability each = 49% end-to-end. Your agents pass. Your system is a coin flip.

Check yours in 60 seconds, free, no LLM calls:
pip install swarm-test

[Suraj Kumar @SurajKumar__10]

@twetsfyp Send

[Suraj Kumar @SurajKumar__10]

@13_narcissus @X Thank you 🙏

[Suraj Kumar @SurajKumar__10]

Appreciate it — let's connect 🤝 No rush at all. If it helps when you dig in: the one-line version is it maps your multi-agent setup as a graph and flags structural failures (no-exit loops, single points of failure, unguarded handoffs) before you deploy — no LLM calls, runs in seconds. Happy to answer anything or run it on a workflow if you ever want to see it on real agents.And sorry for the wrong name above .

[Suraj Kumar @SurajKumar__10]

@founder_sohan Thanks! 🙏 It's the problem I kept hitting building multi-agent systems — figured others were too. What are you working on? Always up to trade notes with people building in this space.

[Suraj Kumar @SurajKumar__10]

Hey Daniel, I build swarm-test: open-source reliability testing for multi-agent AI systems.
Most teams test agents one by one. But the failures that kill production live in the wiring — loops with no exit burning tokens, one agent everything secretly depends on, handoffs with no validation. Invisible in per-agent tests.
swarm-test maps your agent topology and catches these before deploy. Static, no LLM calls, runs in ms.
pip install swarm-test
If you're building multi-agent, happy to run it on yours free.

[Suraj Kumar @SurajKumar__10]

Hi Hariprasad — glad to connect. I build swarm-test: open-source reliability testing for multi-agent AI systems.
Most teams test agents one by one. But the failures that kill production live in the wiring — loops with no exit burning tokens, one agent everything secretly depends on, handoffs with no validation. Invisible in per-agent tests.
swarm-test maps your agent topology and catches these before deploy. Static, no LLM calls, runs in ms.
pip install swarm-test
If you're building multi-agent, happy to run it on yours free.

[Suraj Kumar @SurajKumar__10]

Hi Ritesh — Users are the bottleneck on my startup.
while everyone buy in building multi agent workflows automating repetitive task ,I build swarm-test: open-source reliability testing for multi-agent AI systems.
Most teams test agents one by one. But the failures that kill production live in the wiring — loops with no exit burning tokens, one agent everything secretly depends on, handoffs with no validation. Invisible in per-agent tests.
swarm-test maps your agent topology and catches these before deploy. Static, no LLM calls, runs in ms.
pip install swarm-test
If anyone building multi-agent, happy to run it on yours free.

[Suraj Kumar @SurajKumar__10]

Hi Dev — glad to connect. I build swarm-test: open-source reliability testing for multi-agent AI systems.
Most teams test agents one by one. But the failures that kill production live in the wiring — loops with no exit burning tokens, one agent everything secretly depends on, handoffs with no validation. Invisible in per-agent tests.
swarm-test maps your agent topology and catches these before deploy. Static, no LLM calls, runs in ms.
pip install swarm-test
If you're building multi-agent, happy to run it on yours free.

[Suraj Kumar @SurajKumar__10]

Hi Archaneya — glad to connect. I build swarm-test: open-source reliability testing for multi-agent AI systems.
Most teams test agents one by one. But the failures that kill production live in the wiring — loops with no exit burning tokens, one agent everything secretly depends on, handoffs with no validation. Invisible in per-agent tests.
swarm-test maps your agent topology and catches these before deploy. Static, no LLM calls, runs in ms.
pip install swarm-test
If you're building multi-agent, happy to run it on yours free.

[Suraj Kumar @SurajKumar__10]

Hi Shibesh — glad to connect. I build swarm-test: open-source reliability testing for multi-agent AI systems.
Most teams test agents one by one. But the failures that kill production live in the wiring — loops with no exit burning tokens, one agent everything secretly depends on, handoffs with no validation. Invisible in per-agent tests.
swarm-test maps your agent topology and catches these before deploy. Static, no LLM calls, runs in ms.
pip install swarm-test
If you're building multi-agent, happy to run it on yours free.

[Suraj Kumar @SurajKumar__10]

Hi Blinus �� glad to connect. I build swarm-test: open-source reliability testing for multi-agent AI systems.
Most teams test agents one by one. But the failures that kill production live in the wiring — loops with no exit burning tokens, one agent everything secretly depends on, handoffs with no validation. Invisible in per-agent tests.
swarm-test maps your agent topology and catches these before deploy. Static, no LLM calls, runs in ms.
pip install swarm-test
If you're building multi-agent, happy to run it on yours free.

[Suraj Kumar @SurajKumar__10]

Hi George — glad to connect. I build swarm-test: open-source reliability testing for multi-agent AI systems.
Most teams test agents one by one. But the failures that kill production live in the wiring — loops with no exit burning tokens, one agent everything secretly depends on, handoffs with no validation. Invisible in per-agent tests.
swarm-test maps your agent topology and catches these before deploy. Static, no LLM calls, runs in ms.
pip install swarm-test
If you're building multi-agent, happy to run it on yours free.

[Suraj Kumar @SurajKumar__10]

Hi Jaya — glad to connect. I build swarm-test: open-source reliability testing for multi-agent AI systems.
Most teams test agents one by one. But the failures that kill production live in the wiring — loops with no exit burning tokens, one agent everything secretly depends on, handoffs with no validation. Invisible in per-agent tests.
swarm-test maps your agent topology and catches these before deploy. Static, no LLM calls, runs in ms.
pip install swarm-test
If you're building multi-agent, happy to run it on yours free.

[Suraj Kumar @SurajKumar__10]

Hi Ksenia — glad to connect. I build swarm-test: open-source reliability testing for multi-agent AI systems.
Most teams test agents one by one. But the failures that kill production live in the wiring — loops with no exit burning tokens, one agent everything secretly depends on, handoffs with no validation. Invisible in per-agent tests.
swarm-test maps your agent topology and catches these before deploy. Static, no LLM calls, runs in ms.
pip install swarm-test
If you're building multi-agent, happy to run it on yours free.

[Suraj Kumar @SurajKumar__10]

@a5577517 Hello👋

[Suraj Kumar @SurajKumar__10]

@Sirkelvinne Hi👋