skills

Break down vague epics, feature ideas, or product concepts into clear, workable requirements with acceptance criteria — acting as a Business Analyst / Product Owner. Use this skill whenever the user says "define requirements", "break this down", "what are the requirements for", "write requirements", "BA this", or describes a feature/epic they want decomposed into actionable items. Also trigger when the user shares a product idea and asks to make it concrete, says "I want to build X", or needs help turning a vague concept into structured deliverables.

Rewrite and optimize prompts for clarity, structure, and effectiveness. Use this skill whenever the user says "optimize this prompt", "improve my prompt", "rewrite this prompt", "make this prompt better", "structure this prompt", or pastes a rough draft of instructions/prompt they want refined. Also trigger when the user says "prompt engineer this", "clean up this prompt", or asks you to turn a vague idea into a proper prompt.

Create a principal-engineer-level implementation plan from a Jira ticket. Use this skill whenever the user provides a Jira ticket URL (e.g., https://your-org.atlassian.net/browse/TICKET-123), a bare ticket ID (e.g., APKNOWLEDG-640), or says "plan this Jira ticket". Fetches the ticket details first (summary, description, acceptance criteria, comments, linked issues), then challenges the approach, interviews the user, launches parallel agents to explore the codebase and gather context, and produces a comprehensive plan with alternative approaches, risk analysis, detailed testing strategy, observability, and deployment steps.

Create a principal-engineer-level implementation plan from a Notion page. Use this skill whenever the user provides a Notion URL (e.g., https://www.notion.so/...), mentions a Notion page or doc, or says "plan this Notion doc". Fetches the page content first, then challenges the approach, interviews the user, launches parallel agents to explore the codebase and gather context, and produces a comprehensive plan with alternative approaches, risk analysis, detailed testing strategy, observability, and deployment steps.

Create a principal-engineer-level implementation plan for a task, feature, or bug fix from a plain text description or conversation context. Use this skill whenever the user says "plan this", "create a plan", "how should we implement this", or wants to think through implementation before writing code — even if they don't explicitly say "plan". If the user provides a Jira ticket or Notion page, use /hnh-plan-jira or /hnh-plan-notion instead — those fetch the ticket details automatically.

Deep-research a person's online presence and produce a signal-driven intelligence brief through a VC lens — deal flow, funding rounds, M&A, portfolio signals, market thesis, competitive intelligence. Maintains an extensible source registry per profile. Use this skill whenever the user mentions watching someone's profile, tracking a person online, researching someone's social presence, wants to know what someone has been up to, asks 'what is [person] doing lately', or provides a person's name with context suggesting they want a profile deep-dive. Also triggers for 'profile watch', 'track [person]', 'research [person]', 'what has [person] been saying', or 'add source to [person]'.

Deep PR review with context gathering, build verification, and principal-engineer-level code review. Use this skill whenever the user shares a GitHub PR URL, mentions reviewing a PR, asks to check/review/audit a pull request, or says "review this PR". Triggers on PR URLs from github.com, PR numbers with repo context, or any request to review code changes in a pull request — even if the user doesn't explicitly say "review".

Wrap up a coding session into a pull request — commit changes, create a branch, push, open a PR with a human-readable description, self-review with /hnh-review-pr, fix findings, and post the outcome. Use this skill whenever the user says "create a PR", "wrap this up", "open a PR", "submit a PR", "make a pull request", "send this for review", or wants to turn their current work into a PR. Also trigger when the user finishes implementing a feature/fix and says something like "let's get this merged", "ready for review", or "push this up". This skill handles the entire PR lifecycle from uncommitted changes to a review-ready pull request.

ZenLabs brand design system — colors, typography, spacing, and logo usage rules for all generated visual outputs (PDFs, presentations, documents, HTML, spreadsheets). Use this skill whenever generating any visual output for ZenLabs: PDFs (reportlab), PPTX (python-pptx), DOCX, HTML pages, styled spreadsheets, or any document where layout, colors, or fonts matter. Also trigger when the user mentions 'brand', 'design guideline', 'ZenLabs style', 'our colors', 'our fonts', or asks for something to 'look professional' or 'match our branding'. This skill should be consulted BEFORE choosing colors, fonts, or layout for any ZenLabs deliverable — even if the user doesn't explicitly mention branding. If you're about to pick a color or font for a ZenLabs document, check this skill first.

Evolve and improve existing skills based on session feedback. Use this skill whenever the user gives feedback about how a skill should behave differently, adds a rule like "always do X before Y", points out a missing step in a workflow, says "remember to...", "you should always...", "never forget to...", or any correction about how an existing skill operates. Also trigger when the user says "evolve skill", "update the skill", "improve this skill", "add this to the skill", or references making a skill better based on what happened during a session. This is NOT for creating new skills from scratch (use hnh-create-skill) or for periodic auditing (use hnh-maintain-skills). This is specifically for incorporating real-time feedback into existing skills so they get smarter over time.

Programmatic canvas toolkit for creating, editing, and refining Excalidraw diagrams via MCP tools with real-time canvas sync. Use when an agent needs to (1) draw or lay out diagrams on a live canvas, (2) iteratively refine diagrams using describe_scene and get_canvas_screenshot to see its own work, (3) export/import .excalidraw files or PNG/SVG images, (4) save/restore canvas snapshots, (5) convert Mermaid to Excalidraw, or (6) perform element-level CRUD, alignment, distribution, grouping, duplication, and locking. Requires a running canvas server (EXPRESS_SERVER_URL, default http://localhost:3000).

Calculate Sangousha (SGS) game session payment splits based on player scores and a total pot. Use this skill whenever the user mentions Sangousha, SGS, "score sgs", game scores with player names and numbers, or wants to split a game pot based on rankings. Also trigger when the user pastes something that looks like "player1: X player2: Y" with a total amount — even if they don't say "Sangousha" explicitly.

Generate a cost-focused AWS infrastructure monitoring report for ZenLabs — covering AWS Cost Explorer (monthly/daily spend by service), EC2 metrics (CPU, credits, network), RDS metrics (CPU, storage, connections, memory), S3 buckets, and EKS cluster status. Use this skill whenever the user asks about infrastructure health, costs, spending, resource utilization, or says things like "how's our infra", "infra report", "cost check", "how much are we spending", "resource usage", or any request to understand AWS costs and server health — even if they don't say "infrastructure" explicitly. Also trigger when the user mentions CPU, memory, disk, budget, or asks about AWS bills.

Back up Claude skills, plans, rules, memory index, and global CLAUDE.md to GitHub — safely removing credentials and device-specific info. Use this skill whenever the user wants to save their Claude config, back up skills, sync to GitHub, preserve their setup, or says "hnh-backup". Strips API tokens, usernames, and device-specific paths before pushing to the backup repo.

Interact with AWS infrastructure — S3 buckets, ECR container registries, RDS databases, CloudWatch logs/metrics, and IAM users/roles via the AWS CLI. Use this skill whenever the user mentions AWS, says "check S3", "show ECR images", "RDS status", "CloudWatch logs", "IAM roles", or any request to manage AWS resources. Also trigger when the user mentions "bucket", "container registry", "database instance" (in AWS context), "cloud logs", or "cloud metrics". Trigger on any AWS service URL (console.aws.amazon.com) or ARN. Even if the user just says "check our infra" or "is our database up", use this skill since the infrastructure runs on AWS.

Package a build session into a shareable "Build with AI" demo document — markdown + screenshots + optional screen recordings — and push to the zenlbs/build-with-ai GitHub repo. Use this skill whenever the user says "document this demo", "wrap this up", "create a demo doc", "package this session", "write up what we built", or asks to create a sharing document about how they used AI to build something. Also trigger when the user mentions "build-with-ai", "demo-1", "demo-2", or any reference to packaging a session for sharing with the team. Even if the user just says "wrap it up" or "share this" at the end of a build session, this skill applies.

Set up CI/CD and deploy ZenLabs services to production — GitHub Actions workflow, Dockerfile, ECR, Helm values, ArgoCD, GitHub secrets, and DNS. Use this skill whenever the user wants to deploy a new service, set up a release pipeline, create a production deployment, or says "release this", "deploy this", "set up CI/CD", or "make this live". Also trigger when the user mentions ArgoCD, ECR, Helm values, or the devops repo in a deployment context.

Interact with Facebook via browser automation using the Chrome Claude Extension — fetch comments, reactions, shares, post content, and perform actions like liking, commenting, and sharing. Use this skill whenever the user mentions Facebook, shares a facebook.com URL, asks about a Facebook post, wants to scrape comments or reactions, or asks to interact with Facebook in any way. Also trigger when the user says "FB", "check this post", "get comments", "how many likes", or references any facebook.com or fb.com link — even if they don't explicitly say "Facebook".

Interact with Discord — read messages, send messages, search, react, pin, manage threads, manage custom emojis, and browse channels and members via the Discord Bot API. Use this skill whenever the user mentions Discord, says "in Discord", "on Discord", "post to Discord", "check Discord", shares a Discord channel/message link, or wants to read/send/search messages in a Discord server. Also trigger when the user mentions "server", "channel", "thread", or "emoji" in a context that implies Discord. This skill is for any Discord server the bot has been invited to.

Capture and organize personal knowledge into a structured, searchable knowledge base — with automatic reconciliation against existing entries. Use this skill whenever the user shares a fact, tip, lesson learned, or piece of information they want to remember — even if they don't say 'remember' or 'save'. Also use when the user says 'add knowledge', 'note this', 'I learned that', or shares a standalone insight. Trigger on any statement that looks like the user is teaching Claude something they want retained, not just asking a question. Also trigger when the user wants to refresh, rescan, or audit their knowledge base — phrases like 'rescan Drive', 'update knowledge from Notion', 'audit my knowledge', 'refresh what we know about ZenLabs', or 'sync knowledge'. When the user asks a domain-specific question, check the knowledge base first before relying on general information.

Generate a CTO-level Work Breakdown Structure (WBS) for software projects — covering all SDLC stages at module level, evaluating multiple technical approaches, and delivering a branded PDF. Use this skill whenever the user wants a WBS, project breakdown, technical scope, development plan, module-level estimation, feasibility analysis, or asks to break down a software project into deliverable modules. Also trigger when the user shares a project brief (text, PDF, or doc) and wants it turned into a structured plan, or mentions 'scope this project', 'break this down', 'what would it take to build', 'estimate this project', 'feasibility check', or 'technical plan'. If the user has a software project idea or brief and wants to understand what to build and how — this is the skill to use.

Debug and monitor Kubernetes (EKS) clusters — check pod status, read logs, inspect events, troubleshoot crashes, and view resource usage via kubectl. Use this skill whenever the user mentions Kubernetes, K8s, EKS, pods, deployments, services, or asks about server health, container status, or application debugging in a cluster context. Also trigger when the user says "is the server up", "check the backend", "why is it crashing", "show me logs", "pod restart", or any production debugging request — since the infrastructure runs on EKS. Trigger on namespace names, deployment names, or any kubectl-related request.

Generate a CEO-level project status report for ZenLabs projects by scanning all project data sources in parallel — backlog spreadsheets, GitHub repos, Google Drive, and Notion. Use this skill whenever the user asks for a project report, status update, progress check, daily standup summary, or asks "how's [project] going?". Also trigger on "report", "update me on", "progress", "what did the team do", "standup", "how are things going with", or any request to summarize project activity — even if they don't say "report" explicitly. Works only with ZenLabs projects that have entries in the knowledge base.

Record screen demos and optimize video output. Use this skill whenever the user wants to record their screen, create a product demo video, capture a walkthrough, make a screencast, produce a screen recording, or show how something works visually. Trigger on 'record', 'screen capture', 'demo video', 'screencast', 'screen recording', 'walkthrough video', 'record a demo', or any request to visually demonstrate on-screen activity — even if the user doesn't say "record screen" explicitly. Also trigger when the user has a large video file (MOV, MP4) they want compressed or optimized. Handles browser-based recordings (Playwright automates and records simultaneously at 30fps) and desktop recordings, then compresses output to keep files small.

Interact with Dify.ai — run workflows, chat with apps, manage knowledge bases, view logs, upload files, and manage conversations via the Dify Service API. Use this skill whenever the user mentions Dify, says "in Dify", "on Dify", "run my Dify workflow", "check Dify logs", shares a Dify app URL, or wants to interact with any Dify application. Also trigger when the user mentions "workflow run", "knowledge base", or "dataset" in a context that implies Dify. Trigger on any request to run an AI workflow, check workflow logs or failures, manage RAG knowledge bases, or chat with a Dify-hosted app — even if the user doesn't explicitly say "Dify" but the context makes it clear they mean their Dify platform.

Interact with Google Drive — list, search, upload, download, move, copy, rename, trash, and organize files and folders via the Google Drive API. Use this skill whenever the user wants to work with Google Drive: uploading files, downloading files, organizing folders, searching for files, moving or copying files between folders, checking file info, or managing file sharing. Trigger on mentions of "Google Drive", "Drive", "upload to Drive", "download from Drive", Drive folder URLs (drive.google.com), or any request to manage files in Google Drive. Also trigger when the user wants to upload a local file to the cloud, download a cloud file, or organize files in folders — even if they don't explicitly say "Google Drive". If the user doesn't specify a project or location, assume it's for ZenLabs (check ~/.claude/knowledge/personal/zenlabs.md for Drive structure and file naming rules). Do NOT trigger for Google Sheets operations (reading/writing cell data, formatting cells) — those belong to the hnh-gg-sheets skill.

Interact with Google Sheets — create, read, write, format, share, and search spreadsheets via the Google Sheets & Drive APIs. Use this skill whenever the user wants to work with Google Sheets: creating a new sheet, reading data from a sheet, writing/updating data, formatting cells, sharing with collaborators, or finding sheets by name. Trigger on mentions of "Google Sheet", "gsheet", "spreadsheet" (when the context is Google Drive, not local .xlsx), sheet URLs (docs.google.com/spreadsheets), or any request to create/read/update a sheet in Google Drive. If the user doesn't specify a project or location, assume it's for ZenLabs (check ~/.claude/knowledge/personal/zenlabs.md for Drive structure and file naming rules). Do NOT trigger for local .xlsx/.csv file operations — those belong to the xlsx skill.

Interact with Notion — read pages, create pages, update properties, append content, query databases, manage records, get database schemas, and search across the workspace. Use this skill whenever the user mentions Notion, says "in Notion", "with Notion", "on Notion", shares a Notion URL (notion.so/..., notion.site/...), asks to create/read/update a Notion page or database, or wants to search for something in their Notion workspace. Also trigger when the user mentions "database" or "records" in a context that implies Notion (e.g., project tracker, task list, CRM in Notion). This skill is for the ZenLabs workspace by default. Do NOT trigger for the /hnh-plan-notion skill which creates implementation plans from Notion pages — that's a different workflow.

Set up or update the user's Claude workspace on any machine. Use this skill whenever the user mentions setting up a new laptop, bootstrapping their dev environment, syncing skills/plans/rules from GitHub, restoring their Claude config on a fresh device, or says "hnh-setup". Handles first-time directory creation with basic rules AND subsequent updates/syncs from the backup repo.

Interact with Google Docs — read content, create documents, append text, find-and-replace, and get document metadata via the Google Docs API. Use this skill whenever the user wants to work with Google Docs: reading a doc's content, creating a new doc, updating or appending text, doing find-and-replace, or getting document info. Trigger on mentions of "Google Doc", "gdoc", "document" (when the context is Google Drive, not local .docx), Google Doc URLs (docs.google.com/document), or any request to create/read/update a document in Google Drive. If the user doesn't specify a project or location, assume it's for ZenLabs (check ~/.claude/knowledge/personal/zenlabs.md for Drive structure and file naming rules). Do NOT trigger for local .docx file operations — those belong to the docx skill. Do NOT trigger for Google Sheets operations — those belong to the hnh-gg-sheets skill.

Retrospective skill maintenance — scan all custom skills for improvement opportunities, detect outdated patterns, find where newer skills can replace raw API calls in older skills, and keep the skill ecosystem consistent and up to date. Use this skill whenever the user says "maintain skills", "review skills", "skill audit", "skill hygiene", "improve our skills", "check skills for updates", or after adding a new skill that wraps an API already used by other skills. Also trigger when the user asks about skill quality, consistency, or cross-references between skills.

Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.

Investigate a Sentry issue by fetching full error context (stacktraces, events, tags, frequency) and exploring the local codebase to understand root cause. Generates a prioritized report with suggested fixes and offers to create a Jira ticket. Use this skill whenever the user shares a Sentry issue URL, mentions a Sentry error, asks to investigate/debug/triage a production error, or wants to understand why something is failing in Sentry — even if they don't say "Sentry" explicitly but share a link from a Sentry domain.