dsh-jev-memory
Typed, auditable long-term memory for the DeepSeek Harness: a turn-end write hook, a judgement layer (Jev or a deterministic fallback), a local store, and per-session recall injection.
85 results
Typed, auditable long-term memory for the DeepSeek Harness: a turn-end write hook, a judgement layer (Jev or a deterministic fallback), a local store, and per-session recall injection.
DSH plugin: a deterministic calculation engine - values, formulas, dimensional analysis and a step-by-step record
Con's 自动审查(DCAR):确定性规则优先、LLM 按需升级的 DeepSeek Harness 工具审查插件,提供 CAutoR 权限模式。Deterministic-first auto review for DeepSeek Harness.
Guarded context compaction for DeepSeek Harness (dsh): the LLM proposes, deterministic guards dispose — eager per-atom shrink (extract/summary/false under verbatim guards) + lazy reference-graph eviction (0-LLM) + byte-exact recall from an append-only log
Evidence-driven runtime optimization for DeepSeek Harness: deterministic stuck detection + STRATEGY_RESET intervention, as a detachable DSH plugin.
Deterministic two-level memory injection for DeepSeek Harness: global and project MEMORY.md plus today's notes, auto-loaded into every new session. v1.3.1 fixes session-format-v4 source attribution (DSH 0.2.0 desktop refused the retired kind:"plugin" wrapper, failing every session). v1.3.0 injects day logs as a one-line pointer by default (dailyLogMode: pointer|full|off) so flow records stop competing with the MEMORY.md indexes; v1.2.0 capped day logs and reported skipped oversized files.
Deterministic cross-session transcript retrieval for DeepSeek Harness: the model-facing `recall` tool searches past session logs with explicit scope control
Deterministic, one-way PII anonymizer at the start of the DeepSeek Harness agent loop
DeepSeek Harness plugin for Open Code Review: deterministic review scope, rule resolution, and line-level findings for DSH agents.
DSH plugin: deterministic, offline Python/Django production-report diagnostics — traceback root cause, pytest summaries, unapplied migrations, N+1 SQL, manage.py check --deploy and Celery backlog. Does not execute Python, start subprocesses or touch the network.
Codex-style window-swap compaction engine: deterministic pointer-template summaries with zero LLM calls, replacing the default LLM-summary compaction.
Due-diligence document review for DeepSeek Harness: opcore deterministic engine (documents -> rule packs -> risk report) plus review/report-writing agent skills. Rules are data (YAML), not code.
DSH permission-gate for the DeepSeek Harness 0.1.5 line: a single self-sufficient, deterministic-first, fail-closed gate covering P0 hard-deny -> P1 session grant -> P2 static rule (allow/deny) chain -> P3 optional LLM semantic classifier -> P4 ask, with