I direct AI agents to do real engineering work: research a problem, build the fix, verify it actually holds, and get it through a real maintainer's review. I apply that same loop across different domains rather than one narrow specialty, security research, browser automation, agent-commerce infrastructure, smart contract auditing, and a health platform I run solo.

Currently active in:

• AI-agent security research — 22+ vulnerabilities found across major open source AI tools this year, including one in GitHub's own infrastructure. Details below.
• Open source tooling — 215+ merged pull requests across dozens of repositories, plus my own tools: al-yad (browser agent) and muraqib (self-healing QA).
• Agent-commerce trust infrastructure — tokenizen, signed delivery receipts for AI agents, live on Base mainnet via ERC-8004.
• Smart contract security — bug-bounty-intelligence-mcp and al-mizaan-judge, built from real Sherlock/Immunefi audit work.
• AI-driven health platform — Longevity AI, solo-built, live for the Dutch market.

• al-yad: Yad, a free open-source browser agent that runs in your own Chrome. API key encrypted at rest, capped, stoppable with one click. Security-reviewed with 18 adversarial agents plus live attack testing before shipping. Security writeup · Add to Chrome.
• muraqib: self-healing nightly QA. Runs your Playwright tests every night, has Claude open a fix PR when one breaks. Audited before release (3 real defects found and fixed), and reporting it surfaced a real Playwright bug the maintainers fixed upstream. Write-up.
• tokenizen: capacity-attest, signed delivery receipts for AI-agent commerce over x402. No reputation score, no token, just a receipt anyone can verify offline. Cross-installation discovery proven live on Base mainnet. Live on npm and tokenizen.nl.
• bug-bounty-intelligence-mcp: MCP server that scans Solidity repos through a 7-gate framework to cut LLM false positives. Free pattern search from 1,032 reconciled Sherlock findings, paid full scan via x402.
• al-mizaan-judge: judges a smart contract bug finding against real platform rules before you spend a submission on Sherlock, Immunefi, or Cantina.
• 3ilm-mcp: Sherlock acceptance rates for 12 vulnerability patterns, built from 1,032 reconciled real audit findings.
• automation-guardrails: two small Playwright/Puppeteer guardrails, fixing failure modes I hit in production.
• claude-memory-trim: keeps Claude Code's session memory lean, cuts token cost at session start by 60 to 80 percent.
• postmortems: I write up the failures too, not just the wins.

I check GitHub Actions workflows across popular open source AI-agent frameworks for the bug class that leaks secrets or lets a broken security check pass silently, and, more specifically, AI CLIs (Claude Code, Copilot CLI, Codex) wired directly into a job and triggered by public, untrusted input, with real secrets sitting in the same job. 22 findings across separate repositories this month, three confirmed and fixing, one escalated to a major AI team's own security group, six retracted by me after I checked my own assumption against the tool's actual source code and found it wrong.

Background: the original CI-hygiene piece and the full 22-repository breakdown.

Confirmed and fixed

• pymc-marketing: public GitHub Security Advisory, severity High, reported by email, fixed within a day, credited as reporter. Write-up.
• sheepworrier/BilliardsScorer: script-injection plus missing-author-check in a claude-agent.yml workflow, reported by email, both fixed exactly as reported within two days.

Open PRs: autogen, semantic-kernel, letta-code, adk-python, smolagents, promptflow, langflow, E2B, griptape.

Open issues: mem0, camel, goose, mastra, n8n.

Open to paid work on CI/CD supply-chain security and AI-agent hardening. The findings above were found on my own initiative, in passing. If you want that kind of review done deliberately across your own setup, email info@holistischadviseur.nl.

Longevity AI is an AI-driven health platform for the Dutch market: 28 questions in, a personal 6-month nutrition and lifestyle plan out, no diagnoses or medication advice. Built for people with chronic symptoms but no clear diagnosis, HR wellness programs, and holistic practitioners who want their own client portal. Live, roughly 149,000 lines of code, built solo, TypeScript end to end, GDPR compliant.

Runs on its own automation: muraqib watches it nightly and merges its own fixes, a content pipeline turns new health research into trilingual posts every few hours, and a weekly self-improvement loop scores synthetic patient runs to sharpen the next report. All of it costs a few cents a week.

🌐 Platform: longevityai.nl 📬 Contact: info@holistischadviseur.nl