17245개의 스킬
Run a comprehensive multi-stage automation audit with parallel agents
Comprehensive guide for CMake configuration, build reliability, and project verification in the VS 2026 environment.
Run tests with smart execution strategies - parallel, sequential, or split (UI sequential + unit parallel)
Commit, push, monitor Vercel preview deployment, and create PR only after user verifies
Check Vercel deployment status and fetch build logs on failure
Orchestrate multi-step, multi-agent workflows with HITL gates.
Deployment workflow patterns for Supabase + Stripe Apps — environment promotion, feature flags, rollback procedures, and CI/CD pipeline stages.
Translate an incoming handoff packet or external instruction into a clear, discovery-phase-ready task. Use at the start of a session when receiving work from another session or collaborator.
Validate and operationalize incoming handoff packets into execution-ready plans by checking completeness, resolving ambiguities, mapping requirements to concrete tasks/files/tests, and producing blocker/risk reports with clear next actions.
Use this skill when the user asks to push to personal repository, push to personal account, push personal repo, push with account switch, or push to javierg1975. Switches GitHub accounts (javierg1975), pushes changes, and switches back to NYU account.
Launch a new mission. Initializes mission state and begins the orchestration workflow.
Validate cross-artifact consistency between spec, plan, and tasks
Use before dispatching any agent team. Defines file ownership, parallel dispatch rules, handoff ACK protocol, and the hard rule that no phase runs as a single generalist when a specialist team applies. Non-compliance blocks the phase from starting.
Unified command for the AI handoff workflow. Auto-detects role and state, then executes the appropriate action.
Use when you have a written implementation plan to execute in a separate session with review checkpoints
Lint, run tests, then create a semantic git commit for staged or unstaged changes
Safely parse and apply `## ARTIFACTS` blocks from planning outputs into repository files with strict path validation and atomic writes. Use when a workflow response contains artifact payloads (for PRD/PSD/master plan/feature plan/PR plan/change-review outputs) that must be written exactly as provided.
Build, tag, and publish a new release with cross-compiled binaries and GitHub release
Backend developer role with Node.js, Java/Spring Boot expertise. Auto-loads dev-test, review-code, plan-execute, db-expert, dev-docs, dev-git, task-complete skills.
逐步執行工程計畫，搭配 human-in-the-loop 審查。 每個 Phase 由對應 skill 連續執行所有內部步驟，完成後展示交付物進入 feedback loop。 使用者審查通過後才推進至下一個 Phase。 支援資料夾結構（plan.md + todo/doing/done/）與 Dependency Graph， 允許多個 AI 平行執行互不依賴的 Phase。 當使用者說「carry on engineering plan」「carry on」「繼續工程計畫」 「執行工程計畫」，或想恢復執行先前擬定的工程計畫時觸發。
Generate ready-to-send prompts that stay aligned with the current Phase 0 doc set, runtime truth, and surviving local skills.
Execute implementation plans in background while Pichu stays responsive. Wraps subagent-driven-development with run_in_background: true.
Use when executing implementation plans with independent tasks in the current session
Use this skill whenever the user needs to build and deploy a Spring Boot 2.x application with Java 8. Triggers on mentions of Spring Boot 2 setup, Java 8 Spring Boot, legacy Spring Boot deployment, or Corretto 8 application setup.
Use this skill whenever the user needs to build and deploy a Spring Boot 4.x application with Java 21. Triggers on mentions of Spring Boot 4 setup, Java 21 Spring Boot, latest Spring Boot deployment, or Spring Boot 4.0.2 application setup.
Auto-inject the OTel Java agent into any Java process via JAVA_TOOL_OPTIONS — the OTel equivalent of Dynatrace's LD_PRELOAD. No startup script changes needed. Downloads the agent JAR, sets system-wide env, and any JVM on the host picks it up automatically. Generic skill — works for any Java application (Spring Boot, Camel, webMethods, etc.). Battle-tested: "Picked up JAVA_TOOL_OPTIONS" confirmed on startup.
Groundcrew task decomposer. Breaks any request into queued groundcrew tasks with auto-chained dependencies. Supports skill invocations (/skill-name args), plain text tasks, and mixed pipelines. Use when: "dispatch", "queue tasks", "break this into steps", or invoking /dispatch.
Autonomous deployment skill for the Jelly-legs AI team. Use when agents need to deploy code to GitHub Pages, trigger GitHub Actions workflows, or release updates without human approval. Handles git operations, workflow triggers, and deployment verification.