Hey everyone! 👋
I'm Tarun, a developer at ProjectDiscovery where we're building an autonomous security agent designed for enterprise security teams—think comprehensive security operations with a strong lean towards offensive security and red teaming.
Currently diving deep into Phoenix to instrument our agent workflows and get better visibility into LLM decision-making across security tasks. I'm particularly interested in:
Tracing complex multi-step agent reasoning chains during security assessments
Evaluating multi step workflows
Optimizing prompt strategies for security-specific operations
I'm also running a side experiment where I'm proxying all my Claude Code interactions through Phoenix to analyze and improve my personal coding workflow—curious to see what patterns emerge from daily usage.
Excited to learn from this community and share what we discover along the way!