I've been building a small CLI that runs inside Claude Code — a pre-commit judgment check for the decisions where gut says "yes" but evidence is thin. Architecture calls, pricing bets, scope commitments, the kinds of things a lot of solo founders make alone without pushback.
It's based on Kahneman-Klein (kind vs wicked environments) and consider-the-opposite — the only debiasing technique with peer-reviewed evidence on experienced professionals. Scores 5 dimensions, runs a structured counter-case, outputs a paste-ready memo for PRs / Slack / decision logs.
Free, MIT: https://github.com/dixalex/decision-linter
The indie pain it's built for: you have no one pushing back when your gut is wrong. 2 min of structured friction before hitting "send" is surprisingly good at catching the ones you'd regret.
Curious if anyone else solos through decisions and wants this kind of resistance — or does the 2-min format feel like too much overhead?