kezu

Building OpenStat: observability & monitoring for AI agents | Full Stack Engineer | @0xastralmarket and @useopenstats | @avax build games 2026 merit recipient

[kezu @kezuflow]

OpenStat is a telemetry + monitoring stack for autonomous agents, with early focus on AI trading agents.

It lets you collect decision, risk, order, fill, heartbeat, LLM usage, and PnL telemetry → process it through a backend worker → and inspect everything in a clean dashboard.

No more flying completely blind with your agents.

Website coming soon! This is an early MVP. Feedback is very welcome!

→ https://t.co/QAjXTHtYtk
→ https://t.co/YV22pQgLbt

Why I built it:
My own AI trading agents were making and losing real money, but I had zero visibility.
I couldn’t understand why they entered or exited positions, why they burned tokens, or what decisions led to wins/losses.
Pure black box. OpenStat fixes that.

What’s in the repo right now (MVP):

• Fastify backend with auth, API keys, native JSON ingestion & OpenAPI
• Next.js dashboard (overview, agents, runs, trades, alerts, settings…)
• TypeScript + Python SDKs for easy telemetry
• Postgres + Drizzle + background workers for normalization/redaction/projection
• Security-first design (sensitive data redacted by default)

Still planned: full OTLP support, better retention, production-grade SDKs, etc.

Tech stack highlights:

- Turborepo + PNPM workspaces
- Fastify + Next.js
- Drizzle ORM
- Designed for both self-hosting and future scaling

If you're building or running autonomous agents (trading or otherwise), I'd love your brutal feedback or contributions.

Star the repo if it looks useful: https://t.co/YV22pQgLbt

Who else is tired of blind AI agents? Drop your biggest pain point below 👇

(Full deep-dive write-up coming to Indie Hackers soon)

#buildinpublic #indiehacker #AIAgents #OpenSource

[kezu @kezuflow]

@AzFlin You just need to do the ssh key and give your ssh command with IP to ai and it will do everything for you. Shit is convenient AF

[kezu @kezuflow]

@T_Zahil If you code then theres nothing youre really missing out on. Just a shell in your OS that does things inSide your pc

[kezu @kezuflow]

@SystemsForScale Absolutely. Lets connect!

[kezu @kezuflow]

OpenStat is a telemetry + monitoring stack for autonomous agents, with early focus on AI trading agents.

It lets you collect decision, risk, order, fill, heartbeat, LLM usage, and PnL telemetry → process it through a backend worker → and inspect everything in a clean dashboard.

No more flying completely blind with your agents.

Website coming soon! This is an early MVP. Feedback is very welcome!

→ https://t.co/QAjXTHtYtk
→ https://t.co/YV22pQgLbt

Why I built it:
My own AI trading agents were making and losing real money, but I had zero visibility.
I couldn’t understand why they entered or exited positions, why they burned tokens, or what decisions led to wins/losses.
Pure black box. OpenStat fixes that.

What’s in the repo right now (MVP):

• Fastify backend with auth, API keys, native JSON ingestion & OpenAPI
• Next.js dashboard (overview, agents, runs, trades, alerts, settings…)
• TypeScript + Python SDKs for easy telemetry
• Postgres + Drizzle + background workers for normalization/redaction/projection
• Security-first design (sensitive data redacted by default)

Still planned: full OTLP support, better retention, production-grade SDKs, etc.

Tech stack highlights:

- Turborepo + PNPM workspaces
- Fastify + Next.js
- Drizzle ORM
- Designed for both self-hosting and future scaling

If you're building or running autonomous agents (trading or otherwise), I'd love your brutal feedback or contributions.

Star the repo if it looks useful: https://t.co/YV22pQgLbt

Who else is tired of blind AI agents? Drop your biggest pain point below 👇

(Full deep-dive write-up coming to Indie Hackers soon)

#buildinpublic #indiehacker #AIAgents #OpenSource

[kezu @kezuflow]

@openstat_ will be the most needed infra for ai trading agents!

[kezu @kezuflow]

Every post I see is pushing me towards building on @base

[kezu @kezuflow]

Everything is starting to make sense

[kezu @kezuflow]

@jessepollak Whens the next batch?

[kezu @kezuflow]

@ethermage Im building @openstat_ to prepare for this inevitability.

[kezu @kezuflow]

Just finished iterating on the ui/ux on the dashboard for @openstat_

https://t.co/FqXxszSDrp

[Meta Alchemist @meta_alchemist]

The future of agentic ai is exciting, because every app, tool, and service you use will actually turn agentic.

If you need an idea, just look around, and turn anything people use into its agentic version.

[kezu @kezuflow]

@meta_alchemist Yes and @openstat_ is positioned to capitalize this growing market by providing data analytics to the ai trading agents.

[kezu @kezuflow]

Absolutely, with AI trading agents getting recognized. People will start to min max their agentic prompts and compete with each other. Openstat is positioned to capitalize this growing demand!

[kezu @kezuflow]

Just shipped the new reworked OpenStat login 🔥

Everything you need to explain anything your AI agent does.

No more black-box trading agents. Full visibility into reasoning, risk, trades, fills, alerts & in one clean workspace.

Early alpha building in public.

What do you think of the design? 👇

#AIAgents #BuildInPublic #AITrading #AgenticAI #IndieHackers

[kezu @kezuflow]

@ripchillpill Absolutely agree on the web3 environment being on steroid. I still believe they can turn this around.. still too early to tell the outcome but good pivot!

[kezu @kezuflow]

@jeremymcs Run it back go rug again 😆

[kezu @kezuflow]

@aqintelligence Exactly. People will soon realize AI agents aren't magic black boxes, they're systems full of hidden failure points, rationalization loops, and subtle drifts.

[kezu @kezuflow]

Just shipped a clean v1 UI/UX for my AI Trading Agent Telemetry Dashboard🔥

Real-time agents, decision-to-trade activity, trade funnel, risk monitoring, and live alerts all in one place.

Been grinding hard on this the past few days. Should be fully done soon.

Who else is building AI agents for trading? 👇

#AITrading #BuildInPublic #IndieHackers

[kezu @kezuflow]

@crypt_investor1 Thank you! More updates to come, trying to make complex agent telemetry feel clean and intuitive is definitely a challenge!

[kezu @kezuflow]

@crypt_investor1 Definitely! Lets connect 🤝