Плагины

Cross-session memory for your dsh agent: remembered, recalled, and re-verified instead of assumed.

SQLite-backed long-term memory for DeepSeek Harness: add, search, update and archive entries with categories and tags, automatic conversation logging, tiered compression, and a project dimension that scopes retrieval to the current workspace plus global facts.

DeepSeek Harness plugin: a model-facing `chrome` tool that drives the installed Google Chrome over CDP — screenshots returned as images, DOM/text/eval, console & network diagnostics, real input events. Runs in the Host, so no per-call approval and no bash sandbox limits.

DSH 试金石：给「自改造」补上评测这一环——用一套金标准用例，把改动前 vs 改动后跑一遍、打分、对比，好就留、不好就撤。

Retrospective agent-session evaluation: deterministic grade cards and cross-session regression diffs from persisted logs — no benchmark runs, no LLM judges.

DeepSeek Harness (dsh) host adapter for anysearch: thin Cordis bundle — hooks layer only, all business logic stays in the anysearch server over 127.0.0.1 HTTP IPC

Legal knowledge-base workbench: hybrid keyword/statistical/vector retrieval over local statute .doc/.docx corpora, faceted filtering, charge-name and civil-cause trees, a document editor, and four AI tools (citation check, contract review, AI analysis, AI drafting).

Measure whether a dsh setup change actually helped: register repeatable cases, run them, and diff the results before and after you change rules, skills, or models.

Deterministic cross-session transcript retrieval: the `recall` tool searches past session logs with explicit scope control.

Model-driven context compaction for DeepSeek Harness. The agent decides when and what to compress, commits durable session-log replacements, and ships reversible tool-result compression with byte-exact hash retrieval.

AlphaRSI — three-role self-improvement loop that evolves the training recipe of alpha sequence models. Operator runs walk-forward, Evaluator grades with OOS RankIC, Refiner proposes hyperparameters via TPE Bayesian search; only strictly-better recipes are kept. Offline one-command run, data bundle shipped with sha1 reconciliation, GPU auto-detect. Ships one skill.

DSH plugin for bounded agent evaluation: compare an original with candidate prompt or supported configuration changes against user-supplied checks, record evidence and costs, and leave adoption to the user.

Deterministic engineering orchestration runtime for DeepSeek Harness: the orbit_controller tool and OrbitService run Commander, Executor and Smart Watchdog children through PLAN, EXECUTE, EVALUATE, bounded correction and runtime recovery, with durable state in <project>/.cx/state.json.

Skill routing for DeepSeek Harness: embeds skills and each user task with a local model, keeps the clearly-relevant skills with a gap-based selection rule, and auto-injects their full bodies into the prompt.

Exa-backed web search provider for the ctx.web seam, with Exa Dynamic Highlights on by default and the Exa-Beta header that setting requires; exposes all eight Exa retrieval types, adds an /exa command to change highlights, search type and result count at runtime, and registers an exa_search tool that takes its own result count.

Layered cross-session memory for DSH that pushes only what changed - the first round injects every active entry, later rounds push only added / updated / removed ones, and nothing at all is injected when nothing changed, with keys that merely repeat an entry id left out of the text so the baseline is about 26% smaller. Facts and decisions carry semantic keys (one active truth per key), the model may only write to an inbox that a human promotes, an entry can carry a review date that raises a one-time reminder when due, and search is relevance-ranked with journal lines down-weighted and Chinese matched by bigram. A sidebar tab tells the lifecycle story instead of listing files - entries are grouped by stage (awaiting confirmation / in use / archived) along four axes (topic - one level, human-assigned, type, tag, date), each group can be ticked for batch promote / withdraw / archive / restore / delete / assign topic, and topics can be renamed store-wide or searched on their own. The archived stage reads in three levels - the category says why an entry left (distilled into the docs / useless for good / not needed right now, the first two being defaults you can add to), below it the entry topic stays collapsed until you click it, and entries with no topic sit right under the category. The tab also shows the global and workspace instruction files plus the current injection size. A zero-dependency CLI ships with it, so the store is usable without DSH.

Adds the KnyazevAI API provider and its DeepSeek, GLM, and MiniMax model catalog to DeepSeek Harness.

几何论（共扼谱几何 CSG）知识库插件 for DeepSeek Harness：以 Markdown 文章为唯一真源，程序自动重建 BM25 索引、类型化主张依赖图与修复工作单并做一致性审计，附 869 条真理层与 SymPy 公式验证。开箱即用的离线知识库（31 个 geo_* 工具），零运行时依赖。Conjugate Spectral Geometry knowledge base plugin for DeepSeek Harness: offline BM25 retrieval, typed

Skill recommender for DeepSeek Harness: browse local Codex/Claude/DSH session records, derive a user profile, and recommend suitable open-source skills with a tunable weighted match index (higher index = higher-relevance only). Tools + Web settings panel.

DSH web plugin: a LiveBench tab in the Trajectory view (right of 对话/轨迹). Run LiveBench evaluations against every model configured in the DeepSeek Harness — pick provider/model, category, task, release and question range from dropdowns, watch progress, and

Cross-session knowledge for DeepSeek Harness: ctx.knowledge service + knowledge_* tools, SQLite FTS5 trigram retrieval (CJK substring) with L1 query expansion. V1.11 contract (write/update/search/list/delete + budget + events + error codes).

Real AST-based scan of installed plugins for malicious-intent patterns (dynamic eval, cross-plugin writes, exfiltration-shaped network calls), with an optional periodic schedule and auto-quarantine on critical findings.

DSH session workflows for packaging verifiable capsules, evaluating reusable Skills, and comparing controlled session trajectories.

Project-level Local Context Engine for DeepSeek Harness: retrieval/compression/prioritization over history, code, files, user preferences and agent trajectories, producing a dynamic context window for any LLM.