memory-summarize

To compress AI memory context, this skill condenses memory archives into structured JSON summaries, preserving essential details while removing redundancy and providing compression metrics for auditing.

Memory summarization is a compression technique that condenses accumulated logs and notes into structured summaries while retaining a defined number of recent memories uncompressed to preserve immediate context.

You should use automated memory compression when memory counts or token estimates exceed defined thresholds, triggering automatically to reduce context size in memory-heavy workflows like long-running coding sessions.

Yes, the memory compression script requires the Bun runtime and uses the BrAIny API at localhost:3000 to execute low-cost AI models that produce JSON output with compression metrics.

Yes, a retention policy keeps a defined number of recent memories uncompressed to preserve immediate context, while applying compression only to older archives to generate concise summaries.

Limitations include dependency on the Bun runtime and a local BrAIny API instance, meaning automated memory summarization cannot execute without these specific environment configurations and local service availability.