JoblessJoe/smart-compaction

Lets the model decide when to compact its own context — with compact_now at a safe point and context_status for real token usage — instead of being cut off mid-task by a fixed threshold.

smart-compaction lets the model decide when to compact its own context instead of being cut off mid-task by a blind token threshold. It provides compact_now, which compacts at a point the model knows is safe — for example right after finishing a todo — and works mid-turn, and context_status, which reports real token usage against the session's actual context window. It also compacts automatically once a pending todo completes and the context is more than half used.

Sessions & Messages ★ 1
View on GitHub ↗

Install

dsh plugin --profile web add smart-compaction

Copied verbatim from the README's install line (which the README also gives in an inline Chinese summary). No configuration is needed and it does not change the dsh core.

Compatibility

A DSH plugin installed into a named profile; no config and no dsh-core changes, per the README.

Details

Alternatives

songoao25/dsh-auto-compact · KitDoesIt/dsh-compaction-instant · YYTbit/dsh-plugin-context-compressor

More plugins in Sessions & Messages

Browse more in Sessions & Messages

Guides for Sessions & Messages plugins