AI & ML

kir-kopylov Bundle

«Сделай из промпта быструю команду», «быстрый вызов skill», «копирую промпт из буфера», «часто запускаю этот prompt», «первым после /». Локальный skill Codex.

cleanexpo Bundle

Analyse project skill gaps, generate new skills, browse skill catalogues, and validate skill health. Triggers on "skill gap", "generate skill", "skill catalogue", "skill health", "missing skills", or when the project needs new agent capabilities.

cleanexpo Skill

Playwright CLI and MCP tools for headless browser automation, E2E testing, screenshots, and web scraping. Use for CI-safe browser operations.

smirnovalex-qa Skill

Триаж входящего потока багов — приоритизация, дедупликация, раздельная оценка severity (техническое влияние) и priority (бизнес-срочность), классификация и рекомендация решения по каждому дефекту. Используй когда просят разобрать/отсортировать/приоритизировать баги, затриажить беклог дефектов, решить какие баги брать первыми, найти дубликаты багов, определить severity vs priority конкретного бага, или расчистить очередь входящих дефектов перед планированием спринта. Работает с любым трекером (Jira/YouTrack/GitHub Issues/Linear) через доступный MCP-инструмент или вставленный список. Это НЕ то же, что `bug-report-verify` (тот адверсариально проверяет один репорт на реальность) и не `bug-report-write` (тот оформляет один новый репорт) — здесь речь о разборе и приоритизации ПОТОКА уже заведённых багов.

smirnovalex-qa Skill

Full review of a new feature or branch before merge/release — verification against requirements (YouTrack/a requirements file, if any; otherwise the agent reconstructs the scope of changes from the git diff itself), code review, hunting for inconsistencies in use cases, checking adjacent modules for regression, and live UI testing in the browser, with a final prod-ready verdict. Use whenever asked to check a feature/branch/PR for readiness to merge or release, review/test new functionality, walk through use cases live, check adjacent screens for regression, or do QA/acceptance before prod — even if there is no explicit reference to requirements/an issue in the request.

smirnovalex-qa Skill

Adversarially checks the validity of a bug report — not the formatting, but the problem itself against primary sources (code, spec, git history), trying to disprove the report before confirming it. Use when asked to check/validate/confirm a bug report, to doubt a bug report, or before starting work on a filed bug to make sure it is real and not an agent's hallucination.

smirnovalex-qa Skill

Root cause analysis (RCA) of a defect, incident or failed test based on facts — proving the cause with code/logs/reproduction rather than guessing, separating the proximate and the root cause, using 5 Whys and Ishikawa/fishbone techniques, localizing the introducing commit via git bisect and a separate examination of "why the tests did not catch it". Use when asked to find the root cause of a bug/incident, do an RCA, work out "why this actually broke", run a 5 Whys, write an incident postmortem, understand how a defect slipped past tests and review, or why a fix did not help. Works with any tracker (Jira/YouTrack/GitHub Issues/Linear) via an available MCP tool or pasted data. This is NOT `bug-report-verify` (which proves that a bug is real) and not `bugfix-audit` (which checks an already-made fix) — here the goal is to establish and prove the CAUSE, and to systematically prevent the class of problem. Trigger even without the word "RCA", for example "why could this even happen", "dig down to the root", "how di

smirnovalex-qa Skill

Анализ первопричины (RCA) дефекта, инцидента или упавшего теста по фактам — доказывая причину кодом/логами/воспроизведением, а не угадывая, с разделением непосредственной и корневой причины, техниками 5 Whys и Ishikawa/fishbone, локализацией вводящего коммита через git bisect и отдельным разбором «почему это не поймали тесты». Используй когда просят найти первопричину бага/инцидента, сделать RCA, разобрать «почему это сломалось на самом деле», провести 5 почему, написать постмортем по инциденту, понять как дефект прошёл мимо тестов и ревью, или почему фикс не помог. Работает с любым трекером (Jira/YouTrack/GitHub Issues/Linear) через доступный MCP-инструмент или вставленные данные. Это НЕ `bug-report-verify` (тот доказывает, что баг реален) и не `bugfix-audit` (тот проверяет уже сделанный фикс) — здесь цель установить и доказать ПРИЧИНУ, а также системно предотвратить класс проблемы. Срабатывай даже без слова «RCA», например «почему вообще это могло произойти», «докопайся до корня», «как такое утекло в прод».

lhh666-6 Bundle

Use when you need to create an academic Beamer presentation with original theme and multi-agent review.

lhh666-6 Bundle

Research-domain router that writes `.coord/multi_ai_plan.md` when a single round of work will need two or more delegates AND the work touches research-hub artifacts (`.research/`, `.paper/`, Zotero/Obsidian/NotebookLM pipelines). For a single delegate, use `codex-delegate` or `gemini-delegate` directly — do not invoke this skill. For generic non-research multi-agent decomposition (pure code refactor, generic translation, no research-hub artifact), use `agent-collab-workspace:agent-task-splitter` instead (different artifact `.coord/plan.yml`, different scope). The router decides task splitting, dependency ordering, and reconciliation; the leaves execute.

odjaramillo Bundle

AI/ML development patterns and best practices. Trigger: When working with AI/ML development or model training.

odjaramillo Bundle

Go best practices, effective patterns, and API development. Trigger: When writing Go code, APIs, or microservices.

octagonai Bundle

Generate deep research reports on prediction market events using the Octagon Prediction Markets Agent. Combines real-time Kalshi market data with AI-driven analysis to surface price drivers, compare market vs. model probabilities, and identify potential mispricings across 120+ active markets.

h4vzz Skill

Train machine learning models end-to-end, covering data loading, preprocessing, architecture selection, training loops, validation, and checkpointing.

h4vzz Skill

Injects contextual information into prompts using structured templates to improve AI model accuracy, grounding, and task performance.

h4vzz Skill

Retrieves and synthesizes relevant information from a knowledge base using RAG techniques to provide grounded context for a given query.

h4vzz Skill

Optimize machine learning model hyperparameters using grid search, random search, Bayesian optimization, and Hyperband to maximize model performance within a compute budget.

h4vzz Skill

Compresses context to fit within model token limits while preserving the most critical information for the task at hand.

h4vzz Skill

Optimizes the context provided to an AI model by deduplicating, filtering, reordering, and scoring information to maximize relevance and token efficiency.

yogsoth-ai Skill

Validate causal model consistency

yogsoth-ai Skill

SOP for generating a causal model validation report — summarize coverage, confidence, gaps, contradictions.

yogsoth-ai Skill

SOP for finding gaps in the causal model — missing variables, unexplained effects, weak links.

yogsoth-ai Skill

SOP for flagging contradictions in the causal model — identify conflicting evidence or mechanism claims.

yogsoth-ai Skill

SOP for creating a variable page in the causal model — documents a measurable quantity with its properties.

ske-labs Skill

Build source-dated, deduplicated sentiment evidence or a calibrated sentiment model. Use when separating facts, interpretations, claims, novelty, horizon, and uncertainty across filings, news, analyst commentary, or social sources.

internscience Skill

Use ESMFold model to predict 3D structure of the input protein sequence.

internscience Skill

All tools utilized within molclaw skills connect via the MCP protocol. This skill serves as a unified guide for using the MCP Server. This skill must be loaded to create the MCP server before invoking any tools.

internscience Skill

Predict protein structures with Chai-1 from sequence or FASTA input and return model scoring summaries.

internscience Skill

Implement data transmission between the local computer and the MCP Server using Base64 encoding

internscience Bundle

Execution-ready protein-ligand MM/GB(PB)SA workflow with explicit MCP handoffs and optional analysis.

internscience Bundle

Execution-ready protein-protein MM/GB(PB)SA workflow with MCP-exposed tool names, strict file validation, and failure guards.

encod3d-sec Skill

LLM / AI application attack hunting - prompt injection (direct + indirect), excessive agency, insecure output handling, system-prompt + data leakage. OWASP LLM Top 10. Wiki-first, FIND schema output.

encod3d-sec Skill

MCP server attack hunting - tool poisoning, indirect prompt injection via tool output, rug-pull updates, cross-tool shadowing, over-permissioned/excessive-agency tools, lethal trifecta. Wiki-first, FIND schema output.

encod3d-sec Skill

Drive Burp Suite over its MCP server as an AI triage + attack layer - review proxy history for signals, replay via Repeater/send, OOB-gate blind bugs with Collaborator, fuzz via Intruder (RoE-safe), then hand off to the matching vuln-class hunt. Wiki-first, FIND schema output.

encod3d-sec Skill

Wiki-first "what do I use" lookup - pick the automated TOOL (wiki/tools/), then the PAYLOAD/technique (wiki/payloads/ + wiki/cheatsheets/), for a surface/service/vuln-class BEFORE hand-rolling or working from memory. Use for "tool for <service>", "automated tools for web/<service>", "what should I run on <surface>", "which tool for <X>", "payloads for <X>", "payload arsenal", "cheatsheet for <X>", "how do I exploit <tech/class>", "exploit/attack chain for <X>", "arsenal", any SSRF/XSS/SQLi/SSTI/LFI/JWT/XXE/IDOR/NoSQL/deserialization/CSRF/CORS/CRLF/GraphQL/smuggling/web-cache/OAuth/SAML/MFA/crypto/LDAP/XPath/WebAuthn/file-upload/IMDS/prompt-injection/Modbus ask, plus "privesc arsenal", "CVE arsenal", "default creds", "nuclei templates", "sqlmap/hydra/nmap/bloodhound", "password attacks".

encod3d-sec Skill

Autonomous sub-agent hand-off for a fiddly, fully-specified exploit-compile / escalation RUN - the main agent stays on strategy and the board while a cheap sub-agent runs an exact copy-paste checklist behind a false-root/hostname guardrail. Use for "delegate", "offload", "hand this to a sub-agent", "spin a haiku", or the moment a foothold plus a working escalation vector is identified. Main agent dispatches, waits (no parallel duplicate), integrates the result.