Monglong, known on GitHub as MongLong0214, develops tools for AI coding workflows. CommitLore preserves the reasoning behind repository decisions in Git, while other projects address model routing and agent control of music-production software.[1][2][3]
Start with CommitLore – Git memory for decisions coding agents keep forgetting →
GitHub 
X Career
Contributions
CommitLore stores constraints, rejected alternatives and warnings in Git trailers or notes. It checks proposed records against a session and staged diff, then selects active decisions relevant to the files an agent will edit. Git remains the record of authority; a local database is a rebuildable index.[2]
Ideas
The project distinguishes active, superseded and expired decisions. It also separates records delivered as claims from stronger signed directives. This design addresses a failure of generic retrieval: finding related text does not establish that the repository still endorses it. Its documentation describes integration-dependent capture and does not certify automatic recording on every commit.[2]
Tools, writing & talks
Writing, research & talks
Start here
Git-native decision records with lifecycle tracking and path-scoped delivery to coding agents.
Start with the example context output and lifecycle explanation. Distinguish a delivered claim from an authorized directive, then check what your agent integration actually automates.
Developer and maintainer[1][2]Original source