Skip to content

Agent Conversations

A curated selection of real conversations with AI coding agents like Claude Code, OpenCode, and Codex CLI. Not every session makes this list. These are the ones where i've felt like it's genuinely done something impressive or extremely long running tasks. LLM's are truly powerful with the proper harness, environment and tools around it.

Each entry is a condensed, human-readable retelling of the session. Sensitive details (credentials, internal URLs, proprietary logic) are redacted. The focus is on the technical approach and what made it work.

TitleTokensDate
OpenCodeDocker Swarm Migration with Live VM Testing7.8M
Claude CodeAutomating BetterStack Cleanup via Chrome DevTools5.8M
Claude CodeDebugging a Production Grafana Outage via Browser MCP17M
Claude CodePerformance Profiling a Website via Chrome DevTools MCP10.4M
Claude CodeAgent Team for Notion-Tier Onboarding Polish15M
Claude CodeBuilding an Autonomous On-Call Agent with LangGraph and MCP8M
Claude CodeStreaming Architecture: An 18MB Debug Session18M
Claude CodeTesting an Entire Onboarding Flow via Chrome DevTools MCP19.6M
Claude CodeSelf-Analyzing Conversations to Create a Reusable Skill2.4M
Codex CLIA 13-Agent Swarm for Mobile UI Migration63.3M
Codex CLIA 42-Million Token Design Marathon41.9M
Codex CLIOvernight Autonomous Execution: Video Rebuild While Sleeping10.3M
Codex CLIProduction CI/CD Deploy Failure: SSH Credential Boundary1.7M
OpenCodeDebugging a Circular Import Chain Across 6 Python Packages3.7M
OpenCodeBuilding a 26-Subagent Marketplace in One Session4.8M
OpenCodeResolving 41 Code Review Comments in a Single Session3.2M
OpenCodeMCP OAuth Spec Audit: 342 Messages of Protocol Compliance4.9M
Hello, World