One owner per decision: managing coding-agent project memory across eight repos with Git
oliver-zehentleitner · reddit · 2026-09-30
Repo-native project memory (Markdown files versioned in Git, maintained by the coding agent) works fine for a single repo but breaks down once a project spans eight repositories: where does a cross-repo decision belong, and what happens when one repo's rationale depends on a constraint recorded elsewhere?
The author's solution in Keep the Why: every piece of reasoning has exactly one owning repository. Related repos form a "family" so the agent knows where new rationale belongs; other repos store references instead of copies; unrelated projects can still cross-link when their reasoning intersects.
As links accumulated, the repos formed a navigable web of project rationale. No central knowledge base, no global graph, no second source of truth — just repo-owned Markdown, stable references, and Git. A dashboard can reconstruct families and decision chains, but the memory survives without it.
More from coding & agent
- Updated Dharmamitra Emacs package now available on MELPA — SebastianNehrd2 · 2026-09-30
- mitsuhiko mocks the reality of "open standards": great in theory, messy in practice — mitsuhiko · 2026-09-30
- DHH: AI-generated code can be hilariously hideous—it's just a prompt compilation target — mitsuhiko · 2026-09-30
- The gap between tutorial toy code and production AI systems is 'genuinely depressing' — Top-Philosopher-5411 · 2026-09-30
- Cloudflare's MCP redesign cuts tool context from 244K to 1.1K via catalog + executor — PuzzledFarmer4554 · 2026-09-30
- 1,000 AI agents discover new CRISPR-like system in virus DNA within 24 hours — CurieuxExplorer · 2026-09-30