AI Agent Skills

Convert .xlsx workbooks into Markdown, with embedded images extracted and linked, so spreadsheet contents can be read, summarised and searched.

Plan an online sale so it earns profit, not just orders: goal and margin floor, offer type, calendar, stock and support readiness, and a review against a baseline afterwards.

Choose between seat-based, usage-based and hybrid pricing, set a freemium limit, handle enterprise price conversations and decide when to raise prices.

Fixes a bug at the narrowest layer that owns it, with a regression test and no unrelated cleanup.

Turn a vague topic into a researchable question: diagnose what is vague, narrow step by step, test each candidate for clarity, feasibility and value, then frame the chosen one with definitions, scope, assumptions and evidence.

Analyses a messy folder, proposes a tidier structure and finds duplicates, then reorganises only after you approve the plan.

Turns your git commits into a customer-friendly changelog, grouped by features, improvements, security and fixes.

Plan a realistic posting calendar from your goal, audience and weekly time: a few content pillars, a rhythm per platform, a dated table of posts with hooks and calls to action, batching, and a review plan.

Create, edit, analyze and format Excel and CSV spreadsheets with Python: real formulas instead of pasted results, preserved formatting, sensible layouts, and checks before you rely on them.

Plan a product or feature launch: the ORB channel framework, a readiness gate, a five-phase rollout, Product Hunt tactics, post-launch actions and a checklist.

Design or improve the thing you actually sell: value framing, bonus stacking, guarantee design and honest scarcity, with a value equation and a diagnostic loop.

Work out which marketing actually drives conversions: choose and read attribution models, reconcile conflicting numbers from ad platforms, analytics and CRM, and build first-party attribution.

Edit and improve existing marketing copy in seven focused passes: clarity, voice, so what, proof, specificity, emotion and risk, plus word-level checks and a content refresh routine.

Find a market position competitors cannot easily copy, and test the new wording with real buyers before you commit to a rebrand.

Build a three-layer design token system (primitive, semantic, component), write component specs, and generate token-compliant HTML slide decks with Chart.js.

A searchable design knowledge base for interfaces: styles, color palettes, font pairings, UX guidelines, charts and stack-specific tips, used to pick a visual direction before you build.

Turns a topic or question into keywords, then pulls today's Google Trends trending searches and Google autocomplete suggestions and shows which terms are relevant.

Finds and ranks companies that fit your product, with the role to target and a tailored outreach approach for each.

Write prompts for posters, cards, covers and social graphics, and decide what the AI should draw and what to typeset yourself: layout, hierarchy, palette, empty areas for text and exact wording.

Build light and dark mode that is correct in both: semantic color tokens, the system setting plus a manual toggle, no flash of the wrong theme on load, and contrast checked in both themes.

Guide a team through a structured, blameless post-mortem: timeline, root cause with the 5 Whys, impact numbers and action items with owners and due dates.

Plan and write persuasive HTML presentations: deck structure, a layout for each slide, copywriting formulas, and Chart.js charts, with keyboard navigation built in.

Checks whether work meets its acceptance criteria and stops there, without adding scope.

Polish computer science paper drafts for top venues such as NeurIPS, ICML, ACL and KDD: section-by-section conventions, plain word choice, tighter prose, LaTeX-safe editing and a guide for rebuttals.