An opinionated decision log for AI-assisted development. One CLI command per architectural choice. Branch-aware by default. AGENTS.md canonical. Plugs into the agents you already use — Claude Code, Cursor, Codex, Cline, Continue — with no per-tool drift.
Decisions on a feature branch land in docs/decisions-branches/<branch>.md. On PR merge, an action appends a one-line summary to docs/decisions.md with the PR link. No more conflicts on the shared file.
A single canonical instruction file. CLAUDE.md, .cursorrules, .windsurfrules, .clinerules, AGENTS.md — nine of them become two-line stubs pointing at the canonical source. No drift, no duplication.
A GitHub Action ships with logmind init: every relative [link](path.md) between docs is verified on every PR. A rename can never silently orphan an agent's context again.
Three channels. The same package. Pick whichever lives closest to your other dev tools.
$ pip install logmind$ brew tap thrillmot/logmind && brew install logmind$ npx skills add -g thrillmot/logmind-skillThe agent skill is optional but recommended — it teaches Claude Code, Cursor, Codex et al. when and how to call logmind log in any project that has logmind installed.
Init once per repo. Then log every meaningful choice. Branch routing happens automatically.
$ logmind init
$ git checkout -b feat/auth
$ logmind log "JWT for stateless API auth" \
-r "horizontal scaling without session store" \
-a "server sessions in Redis" \
-i "rotate signing keys quarterly"
✓ wrote docs/decisions-branches/feat__auth.md“The why behind the code disappears faster than the what. logmind makes that disappearance expensive on purpose.”