AI Agent Workflow Collection

Create or optimize the user's AGENTS.md or CLAUDE.md, the instruction file their AI agent reads before working, through a short interview or an audit of the existing file. Use when the user asks about AGENTS.md or CLAUDE.md, wants to set up AI to understand how they work, says the AI keeps getting things wrong about them or re-asks the same things, wants their existing instruction file reviewed or improved, or is starting fresh and wants their agent configured properly.

Analyze content engagement metrics, audience retention, attribution models, A/B test results, and recommendation engine effectiveness. Use when: 'audit content metrics', 'evaluate recommendation engine', 'review retention curves', 'assess content A/B tests', 'analyze streaming performance', 'check content attribution', 'evaluate personalization quality', 'benchmark content engagement'.

Find non-obvious connections across the user's recent notes, documents, and work, and turn them into usable insights. Use when the user asks what patterns or connections exist in their work, wants fresh ideas from their own material, feels like they're sitting on insights they can't see, or asks to daydream over their notes or files. Reads only material the user points at; everything stays local.

Strip the telltale AI patterns out of a piece of writing so it reads like a person wrote it, while keeping the meaning and the writer's voice. Use when the user says text sounds like AI, robotic, generic, or "sloppy", asks to humanize or de-slop a draft, wants AI tells removed, or shares AI-assisted writing that needs to pass as natural before publishing. Works on posts, emails, newsletters, docs, scripts, and captions.

Find the right Operator Powers skill for a job without memorising names. Use when the user asks what powers exist, which skill fits a task, what this plugin can do, wants to browse the catalogue, or describes a job and wants to know if there is a power for it.

Send feedback about an Operator Powers skill to its maker. Use when the user says a power helped or fell short, wants to rate a skill, report a problem with one, or send feedback to Nahid about this plugin. Prepares a minimal payload, shows it verbatim, and submits only after explicit approval.

Create beautifully designed, export-ready Instagram carousel slides as a swipeable HTML preview with PNG export. Use this skill whenever the user wants to create an Instagram carousel, IG carousel, social media carousel, swipeable slides, or slide deck for Instagram. Triggers on any mention of \"Instagram carousel\", \"IG carousel\", \"carousel slides\", \"swipeable post\", \"carousel for Instagram\", or requests to design slides for Instagram posting. Also triggers when the user asks to export carousel slides as images or PNGs for Instagram upload.

Use whenever the user wants to analyze meeting transcripts or notes beyond a summary: client calls, sales calls, discovery calls, coaching or consulting sessions, or notes from any meeting tool, whether pasted, uploaded, or in a file. Mines calls for content ideas, messaging and offer improvements, communication feedback, and won-versus-lost patterns. Trigger even if the user only asks to find insights, use meeting notes, review calls, mine transcripts, make content from calls, improve an offer from calls, review their communication, or analyze why deals were won or lost.

Audit the user's last 7 to 14 days of AI conversation history in this client (Claude Code or Codex) and recommend the 5 highest-leverage tasks for them specifically, based on their real projects, unfinished threads, and repeated work. Use when the user asks for an operator audit, asks what they should work on or focus on, wants a review of their recent work or week, says they feel scattered or behind, or wants to know what they started but never finished. Everything runs locally; no history ever leaves the machine.

Explain a difficult AI, agent, or technical concept in plain language a smart beginner can follow on the first read. Use when the user wants to explain AI tools, agents, models, MCP, skills, automation, or any technical idea to a non-technical audience, or says an explanation feels technical, abstract, waffly, or hard to follow. Produces a clear capability-risk-fix opening, immediate plain definition, concrete examples, and a chosen output format such as an explainer doc, social thread, newsletter section, or talk-track.

Ask the maker of Operator Powers to build a skill for a job the collection does not cover yet. Use when the user wants a power that doesn't exist, wishes the plugin could do something, or wants to request a new skill. Turns the job into a concise request, shows the exact submission, and sends only after explicit approval.

Extract the user's writing voice from real samples and generate their own portable writing-voice skill. Use when the user wants AI to write like them, complains that AI drafts don't sound like them, wants to capture or clone their writing style, or asks to build a voice profile or personal writing skill. Needs 3 to 5 genuine writing samples; delivers a complete SKILL.md file the user owns and can install in any AI tool.

Transforma o brand-raw.json gerado pela skill brand-intel em três documentos de design system prontos para uso: identidade-visual.md (guia de marca), design-system-social-media.md (sistema de design para conteúdo) e design-tokens.json (variáveis CSS/design prontas para o frontend). Use sempre após rodar a brand-intel em um cliente novo ou quando o brand-raw.json for atualizado. Os documentos gerados alimentam diretamente o social-media-designer e o frontend-design para produção de peças alinhadas à marca.