Python development principles and decision-making. Framework selection,

Align MCPAtlas work with roadmap, impact, and success metrics. Use when prioritizing features, defining scope, writing roadmap/ROADMAP.md, or discussing user value and CNCF Sandbox positioning.

Compress long threads into a reusable handoff summary so new chats can continue with minimal context and lower token usage.

Check the current Railway project status for this directory.

사용자의 의도를 객관식 질문으로 명확하게 만들어주는 의도 명확화 스킬.

Plan a sprint with capacity estimation, story selection, dependency mapping, and risk identification. Use when preparing for sprint planning, estimating team capacity, selecting stories, or balancing sprint scope against velocity.

Investigate the codebase and draft a well-structured GitHub issue with evidence; use when asked to file or write an issue.

Validate a GitHub issue against the codebase and expected behavior; use when asked to verify an issue report.

Create thorough implementation plans for complex tasks.

first step of planning

>

Use Anton task-state commands to initialize, validate, and pulse canonical task bundles.

Validar a politica de remotes do repositorio antes de qualquer commit/push; operar em single remote por padrao e usar dual remote apenas quando o PO determinar.

Write or refresh `PLANS.md` for the current task in a way that matches repository policy.

Systematic design quality evaluation. Hierarchy, type, color, space, craft, system. Use when evaluating whether a design is ready to ship, running quality audits, or setting quality standards.

Task sequences, progressive disclosure, and the pacing of an experience over time. Use when designing multi-step flows, evaluating onboarding, or diagnosing why users get lost.

Rank multiple design solutions and explain why. Separate personal preference from quality judgment. Use when comparing design options, evaluating competing approaches, or making design decisions with clear reasoning.

Name what was prioritized, what was sacrificed, and whether the tradeoff was right. Every design decision trades something. Use when evaluating design decisions, defending choices, or helping teams understand what they're giving up.

deal-screener

investment-summary

legal-research

property-report

sop-writer

>

>

Insert handwritten electronic signatures into Word documents (.doc/.docx). Provides a canvas-based signature pad for hand-drawing, parses documents to locate signature fields, inserts signature images, and fills in dates/names. Use when the user mentions electronic signature, handwritten signature, sign a document, e-sign, digital signing, or filling signature fields in Word files.

Monitor and protect your email sender reputation. Use when deliverability drops, investigating spam placement, checking blocklists, or recovering from reputation damage.

Use this skill whenever a user asks for OSINT dorks, Google dorks, GHDB queries, Shodan filters, GitHub code search dorks, search operators, or exposed asset discovery. Prioritize this workflow even if the user does not explicitly mention GHDB.

>

Controller skill for the autopilot: triage, dispatch, integrate, and keep work moving via AgentBus.

executing-plans

Provides the agent with knowledge of the GPU cluster topology and networking configuration for the Vanguard SOC cluster — a live 4-node (expanding to 6) GPU mesh on the 192.168.86.x subnet.

Store new decisions, discoveries, or knowledge for future recall. Use when the user asks to remember something or when significant decisions emerge during a conversation.

Route user requests to the right skill chain with deterministic precedence and minimal overlap. Use when multiple skills could apply, when deciding between rpi/gsd/superpowers/feature-dev, or when a request needs an explicit default workflow and exception rules.

Merge a target branch into the current branch. $ARGUMENTS

Techniques and frameworks for generating new data instances that match the distribution of training data

You have access to a searchable library of tools, skills, commands, agents,

Before doing any work on the user's behalf, you MUST follow this workflow:

Ralph Loop that orchestrates Ralph Loops across a GitHub account/org.

Documentation for GreenLang AI Agent Workforce Deployment

Provides real-time visibility of parallel agent execution within Cursor's terminal

This document defines agent roles, responsibilities, and safe collaboration rules.

Localizados em `~/COMBINED/workspace-config/claude/agents/`:

Scientific computing utilities

The fundamental challenge in any complex problem is combinatorial explosion. The Hubble Space Telescope scheduling problem involves tens of thousands of observation requests, six scientific instrument

Gary Klein and Roberta Calderwood's field research with urban fireground commanders revealed something that challenges the foundation of most AI decision systems: **proficient decision-makers almost n

A fundamental tension in rational agent design: Commitment is computationally necessary (can't constantly reconsider every decision), but blind commitment is irrational (must respond to changing circu

In 1986, Gary Klein and colleagues studied fireground commanders — captains of firefighting companies — to understand how they made life-or-death decisions under conditions of extreme time pressure an

Classical decision theory — and much early AI — assumed that good decision-making follows a recognizable structure: enumerate options, evaluate each against criteria, select the best. This model is cl

When the user wants to plan a posting schedule, create a content calendar, or organize when and what to post. Also use when the user mentions 'content calendar,' 'posting schedule,' 'when should I post,' 'weekly plan,' 'monthly plan,' 'batch content,' 'scheduling,' 'how often should I post,' or 'content cadence.' For deciding what topics to cover, see content-strategy-sms. For writing the actual posts, see post-writer-sms.