---
title: "Tim Kulbaev (@timkulbaev) | Hashnode"
description: "Founder at TMC AI. Building AI automation workflows and open-source MCP servers. Read the latest articles by Tim Kulbaev on Hashnode."
url: "https://hashnode.com/@timkulbaev"
---

@timkulbaev

# Tim Kulbaev

@timkulbaev · Joined March 2026

Founder at TMC AI. Building AI automation workflows and open-source MCP servers.

## Comments

[The shift from AI just finding vulnerabilities to actually triaging them at scale is where the real value is — discovery without prioritization just creates noise. 12 vulnerability patterns is a solid starting set. I'm curious how well the AI handles edge cases where a pattern technically matches but the context makes it a false positive — that's usually where human reviewers still outperform automated tools.](https://hashnode.com/posts/drupal-ai-vulnerability-guardian-triage-12-vulnerability-patterns-at-machine-speed/69ba5d698eaf09883d4dbc66/comment/69ba5f5fc22d3eeb8a5799e4)

Comment · Article · Mar 18 · [Drupal AI Vulnerability Guardian: Triage 12 Vulnerability Patterns at Machine Speed](https://hashnode.com/posts/drupal-ai-vulnerability-guardian-triage-12-vulnerability-patterns-at-machine-speed/69ba5d698eaf09883d4dbc66)

[This is a critical topic as AI agents get more tool access. I build MCP-based automation tools and the credential isolation problem is something I think about constantly — especially when agents can chain tool calls autonomously. The approach of keeping secrets in a separate execution layer that the agent orchestrates but never directly reads feels like the right pattern. Does Aegis handle credential rotation as well, or is that out of scope?](https://hashnode.com/posts/why-your-ai-agent-shouldn-t-see-your-api-keys/69ba117ec22d3eeb8a2525b6/comment/69ba5f0fc22d3eeb8a575e29)

Comment · Article · Mar 18 · [Why Your AI Agent Shouldn't See Your API Keys](https://hashnode.com/posts/why-your-ai-agent-shouldn-t-see-your-api-keys/69ba117ec22d3eeb8a2525b6)

[Really cool use of AI generation combined with FFmpeg for the realtime pipeline — the low-cost angle is what makes this practical and not just a novelty. I work on automation workflows and the idea of chaining AI generation into a continuous streaming output is fascinating. What's the latency like between generating a new segment and it appearing on the livestream?](https://hashnode.com/posts/infinite-piano-stream-how-i-built-a-low-cost-100-ai-generated-youtube-livestream/69ba49d5c22d3eeb8a45624d/comment/69ba5e71c22d3eeb8a56e9b5)

Comment · Article · Mar 18 · [Infinite Piano Stream: How I Built a Low-Cost, 100% AI-Generated YouTube Livestream](https://hashnode.com/posts/infinite-piano-stream-how-i-built-a-low-cost-100-ai-generated-youtube-livestream/69ba49d5c22d3eeb8a45624d)

[I've built similar automation pipelines where the hardest part isn't the technical integration but getting stakeholders to trust the automated output over their familiar spreadsheet workflows. The Airflow + Okta combo is smart for this. Curious whether you're using AI for anomaly detection in the access patterns themselves, or mainly for summarizing review results?](https://hashnode.com/posts/from-spreadsheets-to-automation-rethinking-sox-user-access-reviews-with-airflow-okta-and-ai/69ba2fc0c22d3eeb8a327201/comment/69ba5e19c22d3eeb8a56abea)

Comment · Article · Mar 18 · [From Spreadsheets to Automation: Rethinking SOX User Access Reviews with Airflow, Okta, and AI](https://hashnode.com/posts/from-spreadsheets-to-automation-rethinking-sox-user-access-reviews-with-airflow-okta-and-ai/69ba2fc0c22d3eeb8a327201)

[The invisible Unicode attack vector is really underexplored — most teams I work with focus on sanitizing visible text inputs but completely overlook what's hiding in plain sight in READMEs and docs. The honeypot approach is brilliant for understanding attacker patterns. Have you noticed any differences in how various LLM providers handle these invisible characters in their tokenizers?](https://hashnode.com/posts/i-prompt-injected-my-own-github-readme/69b94a17c22d3eeb8aab61c5/comment/69ba5dcac22d3eeb8a567a42)

Comment · Article · Mar 18 · 1 · [I Prompt Injected My Own GitHub README. Then I Built a Honeypot.](https://hashnode.com/posts/i-prompt-injected-my-own-github-readme/69b94a17c22d3eeb8aab61c5)
