
Project memory for coding agents and humans: the reasoning behind a codebase as Markdown in the repo
Project memory for coding agents and humans: the reasoning behind a codebase as Markdown in the repo
Project memory for coding agents and humans: the reasoning behind a codebase as Markdown in the repo, versioned by Git, so nothing rejected is proposed twice. No database, no daemon, no account.
Missing context-schema backfilled with the current version instead of 0.2.0 (2 of 3 runs in the 0.16.3 series, 1 of 3 in 0.16.2)
Under confirmation-flow: sequential the agent lists every candidate before asking about the first (2 of 3 runs, 0.16.2 series)
When is a context/ topic file too big to be one file? Learn the threshold from real repositories, don't invent one