FinkyAgents
Pre-built AI agents for DevOps & Security teams
Start new thread
Adam Kielar

FinkyAgents - Pre-built AI agents for DevOps & Security teams

byβ€’
Pre-built agents, ready in minutes: Code Review on every PR, Cloud Cost analyst, Repository Security scanner, and an Infrastructure Assistant. Configure where they run and who approves destructive actions. GitHub, Azure, Azure DevOps. Start free.

Add a comment

Replies

Best
Adam Kielar
Maker
πŸ“Œ
Hey Product Hunt πŸ‘‹ I'm Adam, a solo developer based in WrocΕ‚aw, Poland. For the last 7+ years I've worked as a DevOps/SRE/Python engineer at companies ranging from early-stage startups to Fortune 500, and the same problem kept showing up: too many alerts, too many dashboards, too many fragmented tools, and not enough time to turn signal into safe fixes. What's in the box: πŸ€– Code Review Agent β€” reviews every PR through a 3-pass pipeline: 8 parallel scanners (Dockerfile, Kubernetes, GitHub Actions, Terraform, Bicep, container images, secrets, IaC misconfigs) followed by an AI review pass on top. πŸ’° Cloud Cost Agent β€” analyzes Azure spend, flags waste, surfaces optimization opportunities. πŸ”’ Repository Security Agent β€” catches exposed secrets, risky permission changes, supply-chain issues. πŸ’¬ Infrastructure Assistant β€” chat agent that answers questions about your environment and can take action, with destructive operations gated behind human approval cards in Slack. Who it's for: DevOps, SRE, and platform teams who want pre-built agents that work today β€” not "build your own agent with LangGraph" frameworks. Best fit if you're on GitHub, Azure, or Azure DevOps. The story behind it: The first version of FinkyAgents was opinionated and largely autonomous β€” point an agent at your infra, let it run. I shipped that, ran a 50-user closed beta, and the feedback was sharp: nobody trusted opaque agents acting on their cloud accounts. They wanted pre-built agents but with explicit controls. So the product became a catalog: pick an agent, configure where it runs, set who approves what, ship in 10 minutes. What I'd love your feedback on: - Is the catalog model the right shape, or would you rather build your own agents? - Any agent types missing from the catalog you'd want? - The Slack human-in-the-loop approval flow for destructive actions β€” would your team actually use it, or does it feel like friction? For the PH community: I'm opening 50 seats for the next testing cohort today, and PH gets first dibs. Free tier with 25 credits/month, no credit card β€” drop a comment here after signing up and the first 50 are in, with a direct line to me for feedback. I'm hanging out in the comments today β€” happy to answer anything technical or product-shaped.