pmm-hydrate

What problem does it solve?

PMM memory directories can accumulate empty or template-only files when new memory types are added or re-activated, leaving Claude sessions without the historical context needed for reliable recall and reasoning. pmm-hydrate infers content from every populated memory file so new entries start with meaningful context instead of shallow templates, preventing context rot and ensuring every maintain cycle builds on existing history.

Core Features & Use Cases

• Intelligent detection: reads memory/config.md to list active files, strips comments/headings to classify template-only documents, and gates hydration unless --force is specified.
• Flexible modes: operates in single-file, batch (--all), and force scenarios while excluding config.md and secrets.md, always dispatching foreground agents with write permissions and leaving commits to the main context.
• Context-aware synthesis: agents read the template reference definitions plus all populated memory files, synthesize justified content using [system:hydrate] attribution, append or replace files per rules, and return per-file summaries before the host runs git add/commit.

Quick Start

Request pmm-hydrate to populate the target memory file or use --all to hydrate every active template-only file.