10 projects
contextcost
Measure what a repository costs an AI coding agent to read, find what is wasting that budget, and prove the saving by re-measuring.
benchlineage
Local-first provenance, uncertainty, and audit trails for electrical-engineering experiments.
reproweave
Local-first evidence maps and replication plans for engineering and AI papers
willitbreak
Will this dependency upgrade break code you actually wrote? Diffs two versions of a package's API and reports only the changes that hit your call sites.
slowimports
Find out why your Python program is slow to start, and what to do about it.
jsonxray
See what is actually inside a JSON Lines file, including the parts that don't match the rest.
termscope
A live oscilloscope for serial data, in your terminal.
evalint
Audit an LLM eval set the way you would audit an exam: what it can measure, what is dead weight, and which items are broken.
frontiertrials
Private personal benchmarks and rigorous blind studies for AI subscription outputs
runproof
Run coding-agent work unattended in an isolated git worktree, and refuse to accept it unless it passes the checks you declared.
