Showing 9811–9840 of 15737 skills
Guides deciding when a sample needs dynamic analysis, handing it to an off-box sandbox (CAPE, Triage, ANY.RUN, Joe Sandbox) without leaking it, reading the report (process tree, network, drops, config), spotting sandbox evasion, and turning results into IOCs and YARA. Use when static analysis stalls on a packed or evasive sample, or the user asks to detonate a sample or interpret a sandbox report.
rhaist/marq
Guides analysis of a malicious Office or RTF document by extracting VBA macros with olevba, finding the auto-exec trigger, decoding obfuscation, and recovering the dropped payload URL for pivoting. Use when the user has a suspicious .doc/.docm/.xls/.xlsm/.rtf attachment, asks to analyze a phishing document or macro, or needs the second-stage URL from a maldoc.
rhaist/marq
Guides deep static analysis of a PE, ELF or .NET binary without running it, using rabin2 for structure and packing, capa for capabilities, strings, and IOC extraction for pivoting. Use when the user asks to reverse or analyze an executable, DLL or ELF, check if a binary is packed, identify what malware can do, or pull IOCs from a sample. (Quick first look: malware-triage.)
rhaist/marq
Guides safe triage of an unknown sample by identifying the file type, hashing it, checking hashlookup, running quick static analysis, and deciding malicious vs benign without executing it. Use when the user hands over a suspicious file or attachment and asks whether it is malicious, wants a first look at a sample, or needs hashes and a verdict before deeper analysis.
rhaist/marq
Covers defending against phishing, vishing, smishing, pretexting and BEC: out-of-band verification for payments and credentials, caller identity verification, reporting culture, and DMARC/SPF/DKIM and MFA backstops. Use when the user asks how to stop CEO fraud or invoice fraud, verify help-desk callers, prevent deepfake voice scams, or harden against social engineering.
rhaist/marq
Guides running a Business Impact Analysis, deriving RTO/RPO/MTPD, and structuring a business continuity plan aligned to ISO 22301. Use when the user asks to write a BCP, perform a BIA, set recovery objectives, align with ISO 22301, or plan continuity for a cyber outage.
rhaist/marq
Use when starting a new ML task that structurally resembles a known competition (recommendation ↔ retail, demand ↔ supply, churn ↔ fraud, segmentation ↔ classification). Triggers when your task feels "not novel" — there is likely a top solution you can borrow features from. Especially valuable at the start of a project before inventing custom features.
topprismdata/cultivating-ml-agent
Diagnose when ensemble/stacking will fail due to high model correlation. Use when: (1) Stacking or ensemble shows no improvement over single models, (2) Considering weighted average of multiple models, (3) Model predictions are nearly identical, (4) AUC/RMS gains plateau despite complex ensembling. Critical for tabular ML competitions where model diversity determines ensemble success.
topprismdata/cultivating-ml-agent
Use when you have run 3+ feature engineering experiments with no improvement, or when distance to theoretical upper bound is <15pp. Triggers when you suspect "I should stop optimizing features and try something else" — the most common failure mode is wasting weeks past the saturation point.
topprismdata/cultivating-ml-agent
Use when: (1) Considering switching from CPU to GPU for a Kaggle competition, (2) GPU quota is running low and you need to prioritize which competitions deserve GPU time, (3) About to start a model training run and unsure if GPU is worth the 30h/week quota cost, (4) Deciding between CPU AutoGluon vs GPU neural network. Provides a 5-gate assessment framework that produces a verdict: READY (GPU will help significantly), MARGINAL (GPU helps but ROI is low), or SKIP (CPU is sufficient). Inspired by NVIDIA's cupynumeric-migration-readiness, adapted for Kaggle GPU quota management.
topprismdata/cultivating-ml-agent
Use when looking for top-solution tricks on a Kaggle competition, when needing CV-LB gap discussions, when validating a strategy against community experience, or when seeking hyperparameter advice from competition winners. Triggers on phrases like "what did top scorers do", "CV-LB gap in this competition", "winning trick for X", or when starting any Kaggle competition.
topprismdata/cultivating-ml-agent
Optimal blending strategy for Kaggle competitions: 80/20 rule (re-ranking vs raw). Use when: (1) Re-ranking gives unexpected results, (2) Simple average blending underperforms, (3) Single model dominates ensemble. Avoids the trap of 100% re-ranking or equal-weight ensembles. **UPDATED 2026-06-14**: Added asymmetric-blending principle (30/70 rule) for general tabular ensembles. When one model family dominates (e.g., CatBoost in Spaceship Titanic 0.8124 OOF acc), weight it higher than other families. Validated: 30% Silver + 70% Top-5 = 0.8132 OOF (best on SST V3).
topprismdata/cultivating-ml-agent
Match the submission file format to the competition's evaluation metric BEFORE submitting. For ranking-based metrics (AUC, log_loss, MAP, NDCG, RMSLE) you MUST submit continuous probability/score values, not 0/1 class labels or rounded integers. Use when: (1) Preparing the final submission for any Kaggle competition, (2) AutoGluon's default `predict()` returns class labels for classification (must use `predict_proba()`), (3) Unsure whether to threshold predictions, (4) CV score is excellent but LB score is near-random. Real failure case: S6E2 Heart Disease rerun 2026-06-14 — OOF AUC 0.95554 dropped to Public LB 0.88403 with 0/1 submission, recovered to 0.95357 with probability submission.
topprismdata/cultivating-ml-agent
Per-category/per-family models can produce WORSE leaderboard scores than a single global model, even when per-category CV is better. Use when: (1) Considering training separate models per product category/family/store in tabular competitions, (2) Per-category CV improves but LB degrades, (3) Each category has <100K rows from a larger dataset. Covers data volume thresholds, hybrid fallback strategies, and when per-category modeling is appropriate vs counterproductive.
topprismdata/cultivating-ml-agent
Use when: (1) optimizing post-processing thresholds for RMSLE-evaluated competitions, (2) considering zeroing out small predictions, (3) implementing min-sales or adaptive thresholds for time series forecasting, (4) CV improves but LB degrades after changing post-processing, (5) comparing "smart" vs "simple" zeroing strategies. CRITICAL: Never assume that a "smarter" post-processing threshold is better for RMSLE without controlled experiment verification.
topprismdata/cultivating-ml-agent
Preview an iOS app in a live simulator tab and publish it to TestFlight or the App Store with rork-local. Use when the user wants to run rork-local, see their app in the browser-based simulator, capture or frame App Store screenshots, or publish a built .ipa via the asc CLI from localhost.
deonmenezes/mobile-app-skill
Replaces adjectives with evidence, claims with demonstrations, and promises with proof for conversion-oriented writing. Now with mathematical depth metrics, specificity calculus, conversion proof , and verification gates.
Kshitijpalsinghtomar/depth-skills
Re-derives every decision for the mobile context — thumb zones, network reality, attention budgets, and input constraints. Now with mathematical depth metrics, mobile fitness calculus, mobile readiness , and verification gates.
Kshitijpalsinghtomar/depth-skills
Enforces measure-profile-optimize order, preventing intuition-driven optimization of non-bottlenecks. Now with mathematical depth metrics, optimization ROI calculus, optimization , and verification gates.
Kshitijpalsinghtomar/depth-skills
Where am I? Progress, what is stale, and the exact next command to run — start here when unsure
smartworldarafath/Google_Antigravity_Plugins
商业心理与购买阻力底层能力(合并自两个原本总是配对使用的判断层:经济理性层锁定阻力分类,心理感受层翻译成内心独白)。先用六类阻力锁定用户卡在哪(钱/风险/时间/信任/面子/决策成本),再翻译成用户不会说出口的心理独白和内容锚点。处理产品价值、价格、信任、成交、风险、成本、用户为什么不买;区分伪需求和真实付费需求、流量内容和成交内容;落点必须是钱、风险、时间、信任、结果、代价,以及具体的一句话/一个场景/一个动作。
MarchTalk/ai-content-skills
概念核验。两种深度:A 贴身词语核验(写文案时一个词该不该留,三问自检 + 行业黑话豁免);B 深度概念拆解(把一个模糊概念拆到说不清就是没想清——对齐是否各说各话、往回倒查原意、假设检查戳穿伪概念、检验是描述行动还是替代了行动、去词测试、判断是信息缺口还是执行缺口)。别只停在"是不是 AI 空词"就收工,那只是 A 模式;用户要"深度拆解"必须走完 B 模式。
MarchTalk/ai-content-skills
账号定位、人设定位、内容栏目、目标用户、承接路径。先锁定目标用户避免吸引错人,把产品优势翻译成用户场景,区分个人 IP / 创始人 IP / 品牌 IP / 门店 IP / 工厂 IP。内容支柱允许商业向(服务成交)和人设向(服务 IP 建立)并存。输出必须包含人设、目标用户、内容支柱、承接路径。
MarchTalk/ai-content-skills
深挖对话的整理成稿工具。深挖已经完成(不是从零开始写)之后,把对话里的实质内容忠实、完整地整理成正式稿件——不做精选,不设默认字数,保留追问-反驳-修正的张力,张力归属必须忠于事实。 触发:/organize、「整理成稿」「把这个整理一下」「深挖完了帮我整理」「把这段深度拆解的内容整理」 Deep-dig-to-draft organizer. After a free-form dig is done, faithfully and completely organize the dialogue's substantive content into a formal draft — no curation, no default length, preserving argumentative tension with accurate attribution.
MarchTalk/ai-content-skills
抖音、小红书、视频号的内容适配。抖音抓前 3 秒 / 冲突 / 节奏 / 完播 / 评论情绪;小红书抓封面点击 / 标题关键词 / 搜索词 / 前几行 / 收藏理由;视频号靠熟人信任 / 老板人设 / 长期口碑 / 朋友圈联动。同一选题必须能改成不同平台版本。
MarchTalk/ai-content-skills
最终质检。按内容定位(流量 / 信任 / 成交 / 人设 / 思考)分别用对应的质检清单。检查空话、模板腔、AI 腔、重复句式、目标用户错位、是否缺场景 / 代价 / 信任锚 / 承接(成交型)或人设印象(人设型)。给修改意见,必要时直接重写。人设 / 价值观 / 思考型内容不挂商业承接不算缺陷。
MarchTalk/ai-content-skills
选题策略,决定什么值得写。从用户处境、行业矛盾、成交阻力、价格争议、客户误解、人设观察、价值观、长期思考找选题。每个选题必须标注生态资产、定位类型、目标用户、平台、流量理由 / 共鸣理由、承接或人设目的、风险点。承接按内容类型处理,不强制每条围绕变现;认知、观点、故事、人设、价值观、思考类内容可以不挂商业承接。
MarchTalk/ai-content-skills
Move a three.js game or scene into a Unity 6 project - the scene as a GLB with glTFast, a Unity scene built from it with the camera, fog and lights, the game state as C#, and a list of what does not carry over. paprika3d's unity_export tool.
Keeno-Corp/paprika3d
Visual regression, screenshot diff and before-after of a three.js, React Three Fiber or WebGL page. Catch unintended visual changes with paprika3d's look save and compare tools.
Keeno-Corp/paprika3d
Generate scholarly bilingual (English + Traditional Chinese) explanations for classical Chinese verses. Use when explaining Yilin, I Ching, or Tang poetry lines.
augchan42/inkstone