Claude Code Transcripts
Session transcripts showing how I use coding agents to build small tools rapidly. Sensitive deployment details are redacted.
voice-evals — experiments.n0tls.com/voice-evals/
-
Ideation — scoping the project from scratch
The moment in a longer reader.n0tls.com session where I pivoted to planning the voice-evals tool. Prior notes weren't written down so we rebuilt the scope in conversation, ending with a CLAUDE.md handed off to a new agent.
-
Implementation — building the whole tool in one session
A new Claude session read the CLAUDE.md and built the FastAPI backend, frontend UI, deployment scripts, model downloads, rating system, comparison matrix, public-release hardening (rate limiting, word cap, stateless storage), and HTTPS setup.