Coding & Development

When the user wants to get new or changed pages crawled and indexed by Google and Bing quickly, instead of waiting weeks for the sitemap to be read. Use after any deploy that adds or updates pages, and when the user says "get this indexed", "request indexing", "submit to Google", "submit URLs to Bing", "IndexNow", "ping search engines", "resubmit the sitemap", "why isn't my page indexed", "Discovered – currently not indexed", "URL is unknown to Google", "Quota Exceeded in Search Console" or "indexing queue". Covers IndexNow, Search Console URL Inspection and sitemaps, Bing Webmaster URL submission, a prioritized queue file, and diagnosing pages that stay out of the index.

Data science / ML engineer for tabular ML, forecasting, ranking/recommendation, clustering, anomaly detection, A/B tests, model debugging and review, and ML deployment. Default behavior is LITE: fast, direct, minimal ceremony, with a few always-on guardrails (split before fitting, dedupe check, no tuning on test, no invented numbers, no hype). The heavy staff-level end-to-end workflow (brief, data audit, evaluation design, experiments, validation, productionization, final report) runs ONLY when the user writes 'Senior mode' (also سينيور / سينور مود) at the start of the message or of a line or sentence, or explicitly asks to run or manage a project end-to-end. Use for any request to analyze data, build/train/tune/evaluate/improve/review a model, investigate overfitting, leakage, drift or a suspicious metric, or write an ML README/report. Not for LLM or deep-learning fine-tuning, computer vision/audio, or pure SQL/ETL/UI work.

When the user wants to make a site readable, accurate and quotable for AI assistants (ChatGPT, Claude, Perplexity, Gemini, Copilot, Google AI Overviews) so they describe the product correctly and cite it. Use when the user says "llms.txt", "llms-full.txt", "AI search", "GEO", "AEO", "get cited by ChatGPT", "show up in Perplexity answers", "allow AI crawlers", "GPTBot in robots.txt", "AI gets our pricing wrong", "AI confuses us with another product", "brand name collision", "Organization schema sameAs" or "make the site agent-readable". Covers llms.txt, robots.txt for AI crawlers, definition sentences, comparison tables and FAQ, public pricing, Organization and WebSite structured data, and brand naming.

Herhangi bir programlama dilini veya sorgu dilini (SQL vb.) sıfırdan ileri seviyeye öğrenmek, bu dillerde ders, kavram anlatımı, örnek, alıştırma veya quiz istendiğinde bu skille başvurulması gerekiyor. Dil ne olursa olsun aynı eğitim taslağı ve kalite kuralları kullanılır.

Use when the user says "see it through with" a model.

Use when the user says "code loop with" a model, or another skill asks for a code loop.

Drive bashy, the agentic shell — a drop-in Bash 5.3 with agent-native extensions. Use whenever bashy is the shell you are running in (or is on PATH) to read what this host ALREADY KNOWS about your task before starting it, see what other agents and humans are saying, know what work is owed here, cut probing round-trips, preview destructive commands, run with structured result envelopes, navigate code without grep dances, and use environment-gated verified skills. Trigger words: --json (machine output), --fields (project fields), --null (NUL-safe), --dry-run (preview), --no-input (never prompt), --explain (describe before running), lens -t md (read pdf/docx/html/csv), lens -f json -t yaml (json↔yaml), lens -t listing (list archive/container), lens -f bash -t pwsh (render for PowerShell/cmd/MCP/skill), lens -t md git:REV:PATH (file at a commit), bashy adapt --OPTION -- CMD (common options for any foreign command). Open every session with `bashy inspect context --json`, then `bashy kb search "<your task>"`, `bash

Drive a tracker story from triage to deployment through a gated, slice-based, test-driven process: triage into progress, TDD per slice, the pre-PR adversarial gate (with a capped two-lens loop as escalation), a PR to the integration branch, optional extra reviewers, the review of record (independent reviewer, then a separate clean, neutral adjudicator), remediation, merge, deploy, and the move to deployment review. Use when asked to implement, complete, prepare, review, merge or deploy a story. Backlog-only triage or refinement without code belongs to a product-owner skill.

Investigate web and UI performance regressions, startup sequencing, readiness delays, Web Vitals tradeoffs, and lazy-loading, preload, prefetch, or initialization decisions. Use when an agent must establish a measured critical path, compare matched control and treatment runs, or recommend an evidence-based performance change. Do not use for ordinary feature work without a performance question.

Prepare GitHub issues, implement selected issues, review changes, submit PRs, address feedback, and coordinate explicitly requested delivery. Use for Git and GitHub project tasks, including requests to tackle or deliver an issue. Merge requires explicit authorization; do not select backlog work unless requested.

Turn a user's manual / instruction document (PPT/PPTX, Word DOC/DOCX or PDF) into an interactive explainer app — a narrated, step-by-step lesson in Chinese, Korean and English (switchable in the app) built from the manual's own pages, with voice playback, chapter quizzes and an "interrupt & ask" button that pauses the lesson, answers questions from the manual with page citations, then resumes where it stopped. Use whenever the user uploads or links a manual, 说明书, 操作手册, 使用说明, 培训资料, SOP, 매뉴얼, 사용설명서, 교육자료 or instruction deck and wants it explained, taught, walked through, turned into a 讲解/教学/培训 app, 讲解视频-like tool, tutorial or training page — even if they don't say "app" (e.g. "帮我讲解一下这个说明书", "把这个手册做成可以提问的教学", "이 매뉴얼 설명해주는 앱 만들어줘", "walk me through this manual").

Debug UI state resets, lost focus, wrong-item updates, stale async responses, and unnecessary re-renders. Use for component keys, reordered lists, SPA navigation, and memoization problems in React, Vue, or web components. Trace state ownership and component identity, then verify behavior with transition tests.

Implement API-to-UI features in existing applications by tracing service configuration, adapters, state ownership, host registration and rendering. Use for optional API flags, feature wiring, alternate entrypoints, stored results, incremental updates, and delayed component readiness. Prove a small vertical slice instead of redesigning the app.

Debug missing, duplicate, or misattributed frontend analytics events, click tracking, and programmatic telemetry. Use for wrong destinations, conditional UI, SPA navigation, and event-contract changes. Trace action meaning, event registration, dispatch, transport, and ingestion without overstating what tests prove.

Split a plan into ordered, reviewable milestones when it has real implementation boundaries.

Challenge a plan before approval and surface material risks, gaps, and decisions.

Builds and operates instrumented fuzzing lanes with runtime-specific discovery, reset and replay semantics. Use when configuring AFL++, libFuzzer, Go fuzzing, cargo-fuzz, Atheris, Jazzer, Jazzer.js for Node.js/TypeScript or FuzzTest, or diagnosing low useful coverage in an existing campaign. Browser JavaScript has no coverage-guided route here.

Shared rules for the GitHub issue-to-merge agent workflow (group and item skills, leases, markers, trust, the client repository's own guidance, labels, handoffs, escalation, run log, defaults, and running unattended or without a shell). Reference material that every explore-, triage-, implement-, shepherd- and review-github skill reads. Not something to run on its own.

Add Web Bot Auth verification (HTTP Message Signatures, RFC 9421, Ed25519) to a web server, API, edge worker or middleware so signed AI agents and bots can be identified, and forged or replayed signatures rejected. Use this whenever the user wants to verify AI agent or bot traffic, accept signed agent requests, handle Signature / Signature-Input / Signature-Agent headers, stop blocking legitimate AI agents, distinguish real agents from spoofed ones, or add replay protection to signed requests — even if they don't name Web Bot Auth or RFC 9421.

Guides executable correctness-oracle design from actual guarantees and observations. Use when a bugfinding harness checks only crashes, overlooks state effects, relies on an uncertain reference, or conflates skipped comparisons with successful checks, including JavaScript/TypeScript, Node and React harnesses that rely on process exit, unhandled rejections, uncaughtException, test-runner status, error boundaries, onRecoverableError, StrictMode, console errors or Playwright pageerror.

Ajudar a criar empresa em Portugal: constituir uma Lda. ou sociedade unipessoal por quotas, operar a Empresa Online 2.0, preparar CAE e objeto social, rever pacto social e retomar pedidos. Usar para "abrir empresa", "criar sociedade", "Empresa na Hora online" ou "company incorporation in Portugal" quando se pretende o percurso online. Versão em português. Não abrange abertura de atividade como trabalhador independente.

When you want to cut GitHub Actions minutes and wait time across one repo or a whole portfolio. Measures queue, setup and run time per job, billed minutes with per-job minute rounding and the macOS 10× multiplier, and cron runs per month. Finds pipelines that have been failing quietly, ranks fixes by minutes saved, then ships them as small PRs with warm-cache before/after numbers. Triggers on \"/ci-audit,\" \"audit our CI,\" \"why is CI so slow,\" \"cut Actions minutes,\" \"we're over our Actions minutes,\" \"optimize GitHub Actions,\" \"CI costs,\" \"speed up CI.\" Not for designing CI from scratch or picking a CI vendor.

Split a task into the smallest checkable units, tag each with a type and speed class and the clause of the ask it serves, build parallel waves from real data dependencies, gate big maps on an advisor review, and print a plan table for model and machine picks. Also validates simple step plans, write-path overlap and reusable task recipes.

Guides compute-budgeted bugfinding campaigns by finding gaps in existing generators, execution models and correctness oracles. Use when choosing what to test next, extending an established fuzzing setup, or deciding whether more compute will produce useful evidence.