Plugins

Cross-model continuity for DeepSeek Harness: memory, skills and rules live in local files, so judgement survives session and model changes. Not about storing more memory — about using it well after the model changes.

StartLux Decision: a cheap judgement layer for DeepSeek Harness — all nine decision points (tool authorization, model routing, delegation, compaction, turn continuation, retrieval, skill pick, session title, COT translation) are implemented and individual

Code security defense review judge for DeepSeek Harness: detects sessions that are writing code, reviews them against the Code Security Defense Architecture Rules, recommends which security architectures the program needs (one or several), checks conflicts between them, teaches the session AI after repeated failures, and teaches the session AI after repeated failures; once you grant it review authority in its panel, it gates task completion until the review passes (without that authorization it only observes).

Pavement structural dynamics as a judgement manual: 59 cards over 8 chapters flagging which chapter name oversells its contents, which formulas break when copied, and which numbers cannot be cross-quoted.

Pavement design methods as a judgement manual: 76 cards over 9 chapters on method selection, indicator assignment, parameter calibers, axle-load conversion and climate zoning, with the airport chapter recovered by re-OCR of the intact PDF.

Writing council: parallel multi-model drafting, judge consensus & critique synthesis, AI prompt refinement, visual concept illustration, and Markdown file preview.

Judges the working model's own emotion each turn and silently re-injects it before the next turn, closing a self-observation → state carry-over → behavior shift loop. Ships a host plugin plus a composer-dock panel, a read-only emotion_status tool and an optional read-only HTTP route.

Multi-agent Chinese debate simulator with eight debaters, live two-column discussion and draft notes, judges and scoreboard, single-side clocks, per-round ZIP archives, a training side with a configurable end condition (win-rate threshold or fixed rounds), and a Word prep package.

Publish size check skill: verifies a deliverable's geometric specs against a target platform before publishing - pixel dimensions, aspect ratio, dpi, file size, format, physical print width and colour mode. Geometry only: it does not review content compliance (QR codes, text-to-image ratio, copyright and prohibited content are out of scope). Pure Python standard library, no third-party packages, and inputs are never modified. Python 3.8+. Covers 10 platforms and 20 slots with per-platform confidence labels - WeChat MP, Xiaohongshu, Bilibili, Zhihu, Weibo, Douyin, WeChat Channels, journal submission, print and client delivery - and resolves a platform by id, Chinese name or alias with fuzzy matching. Two trigger modes: asked directly, or as a parasitic self-check that runs by itself once an image, poster, chart or export workflow has written a file and the user mentioned a publishing target, reporting in one line when everything passes and only expanding when something is wrong. For journal submission and print it judges by physical width (pixels / dpi), which catches files whose dpi metadata passes but whose actual printed size is too small; for multi-image posts it checks whether the whole set is consistent and names the odd one out rather than listing every file.

Evidence-first reuse skill for DeepSeek Harness: before implementing, porting, choosing dependencies, or troubleshooting a tool, it searches the local project and machine for existing solutions in layers, judges reuse on checkable evidence, refuses reuse when essential requirements or risks are unmet, and reports what could not be verified. Ships as a DSH bundle that registers the skill on ctx.skills.

A social deduction game played against your own model configurations. A table has N seats (six by default, one of them yours) and some of the rest are played by agents bound to different provider, model, temperature, system prompt and disguise settings; you interrogate them and cross-check the answers to work out which seats are not human. Seat identity and each agent model configuration stay hidden in the host state until a seat is judged, then the disguise parameters are revealed. Every round exports a per-seat conversation JSONL plus the judgment record, and the same seed with the same move script replays byte-identical, so runs with different models or disguise strengths can be diffed.

Checks an FMEA worksheet for element completeness and for the arithmetic that follows from it - severity, occurrence and detection against the resulting risk priority number - following published FMEA method standards. Reports inconsistencies; it does not judge whether the ratings themselves are appropriate.

Checks an electrical work permit for the elements it must carry and for the issue and permission sequence the current electrical safety regulations require. Reports the differences it finds; it does not judge whether the work was safe or whether the permit was validly authorised.

Checks a bid evaluation scoring matrix for internal arithmetic consistency: that each factor's maximum, the scores recorded and the total agree with each other. Reports the differences it finds; it does not judge whether the evaluation itself was correct or fair.

Turn a GitHub repository's raw counts into a defensible judgement — contributor concentration and bus factor with single-author files, merge time reported as a median beside the mean so an outlier is visible rather than hidden, and an issue backlog with pull requests and drafts filtered out. Reads the public REST API with no SDK, and never claims a truncated page is the whole answer.

Review code with Open Code Review: deterministic scope and rule resolution, judged by your own DSH model, no API key.

DeepSeek Harness permission guard: work inside the workspace is never interrupted; only the extra permission a call asks for is judged — machine-safe actions pass, system changes ask with a consequence analysis, destructive ones are refused.

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

DeepSeek Harness plugin bundle: grill_tree, a fixed-policy design-space search tool. Expands one decision per node, fans out independent proposers, has a separate judge prune the branches, beam-limits each depth, and ends with complete designs or an infea

DeepSeek Harness plugin for CVE impact retesting: fingerprint a target, judge it against a CVE's affected range, gather passive evidence, and report a verdict with confidence and limitations.

Resume screening plugin for HR in DeepSeek Harness: ingest up to 100k Word/PDF/Excel resumes into one searchable library with skills / coarse categories / ingest time, auto-ingest & de-dupe new resumes, then screen by plain-language criteria — rule pre-filter in the database plus LLM fine-judge on the shortlisted few, CSV export. Works on DeepSeek Harness 0.1.7 through 0.2.x, tested on 0.2.0-rc.1.

Multi-role parallel argumentation for DeepSeek Harness: runs the real Codex and Claude CLIs in parallel, has the DSH model (Judge) write a third-party verdict back into the conversation, plus Obsidian-style single-agent direct chat and per-role model config.

DeepSeek Harness 插件：生成前注入「人味」风格约束 + AI 味检测报告（纯代码、零 token、结果可复现）。dsh plugin: human-flavor style injection + AI-flavor detection report, judged by code.

Multi-agent isolated debate for DeepSeek Harness: debaters of different temperaments research independently (no herding), then cross-examine, then judges converge to a conclusion. Runs as a background job with per-agent progress, token accounting, wave-parallel execution, and session-referenced resume.