The "context" topic on GitHub groups 16 open-source projects in the DeepSeek Harness (DSH) ecosystem, led by dsh-auto-memory with 90 GitHub stars. dsh-auto-memory — Proactive associative memory for DSH: zero-prompt recall injected before the model speaks, three-layer auto-consolidation, skill crystallization, and Astra-style context management - handoff ledgers, PLAN whiteboard, water-level sensing. Every project here is indexed by DSH Universe with live GitHub data — stars, activity and install status — so you can compare and install directly.
Proactive associative memory for DSH: zero-prompt recall injected before the model speaks, three-layer auto-consolidation, skill crystallization, and Astra-style context management - handoff ledgers, PLAN whiteboard, water-level sensing. Local-first, model-agnostic, zero deps. Proactive associative memory + Astra-style context management: auto recall / auto consolidation / skill crystallization / handoff ledger and whiteboard that survive across windows / water-level sensing.
Structured context for DeepSeek Harness: split the conversation into typed sub-contexts (reasoning / tool calls / replies) indexed by a table of contents. dsh structured context plugin: split sub-contexts by type, indexed by a structured table of contents.
Curate tool output before it enters the model's context — keep error/warning lines and head/tail, spill the full text with a pointer. ~24x smaller model view, measured. For DeepSeek Harness.
Context chip for DeepSeek Harness: cost per step, cache-hit, bands, compaction threshold, balance, tariff. Never calls a model. | Индикатор контекста для DeepSeek Harness: цена шага, кэш-хит, полосы, порог компакции, баланс, тариф. Модель не вызывает и токены не тратит.
