AI & ML

khadir-syed Bundle

Diagnose a failing test or reported defect, propose a fix, and apply it only with approval at each stage — investigate, then fix, then verify. Use for the same job as the Test-Fix Loop Agent but with a human approving every step; do not use for a general code review or for a fix already fully specified by the user.

khadir-syed Bundle

Classify an unstructured request as software-bug-shaped, product-launch-shaped, or content-publish-shaped, then hand off to that category's autonomous agent — no pause on the classification itself. Use for the same job as the controlled variant when you want fewer check-ins; do not use if you want to review the classification before anything downstream runs, and do not use for a software-bug-shaped request unless you understand that path completes with zero further approvals — see the Software category below.

khadir-syed Bundle

Classify an unstructured request as software-bug-shaped, product-launch-shaped, or content-publish-shaped, then hand off to the matching controlled agent — stopping for approval on the classification itself before any downstream work begins. Use when you don't yet know which of the three agents applies; do not use once you already know which agent you want, and do not use for requests that are clearly none of the three (general questions, social-media asks).

khadir-syed Bundle

Chain the Requirements Gap Investigator, a launch brief draft, and the Feature Launch Readiness Checklist into one run, pausing only once — right before the single write. Use for a feature that needs to go from a rough ask to a launch-readiness report in one pass; do not use when you want approval at every hand-off — use the controlled variant instead.

khadir-syed Bundle

Chain the Requirements Gap Investigator, a launch brief draft, and the Feature Launch Readiness Checklist, pausing for approval after every step. Use for the same job as the autonomous variant when you want a checkpoint at every hand-off, not just before the write.

khadir-syed Bundle

Chain the Content Brief Gap Investigator, a content draft, and the Content Publish Readiness Checklist into one run, pausing only once — right before the single write. Use for a content request that needs to go from a rough brief to a publish-readiness report in one pass; do not use when you want approval at every hand-off — use the controlled variant instead.

khadir-syed Bundle

Chain the Content Brief Gap Investigator, a content draft, and the Content Publish Readiness Checklist, pausing for approval after every step. Use for the same job as the autonomous variant when you want a checkpoint at every hand-off, not just before the write.

nhouseholder Skill

Session end. Write a lean handoff in this project's handoffs/, commit it on the current branch. Do not auto-ship. Save standing memory to Engram MCP.

twhsi Bundle

Create a personal athlete 81-cell MandalArt grid from an Ohtani Shohei-style 64+8+1 model. Use when the user asks for 大谷翔平 81 宮格, 個人運動員81宮格, sports skill maps, athlete training Mandala charts, badminton 81 grids, or editable JSON/SVG/PNG-ready athlete development templates with Ohtani-style colors.

zengxiansheng007 Bundle

Plan, execute, review, repair, and stabilize Agent-driven Web UI tests using Codex orchestration, Midscene-first exploration, Playwright verification, Solution D evidence, failure attribution, governed execution-experience capture and Playwright/Hybrid regression assets. Use when the user asks to 运行或设计 UI 测试, 使用 Midscene/Playwright 测试真实页面, 分析 UI 测试失败, 生成测试报告, 沉淀或消费 checkpoint/UI 测试经验, 治理语义只读 POST, or 沉淀回归脚本.

zengxiansheng007 Bundle

Stress-test plans, requirements, test designs, implementation choices, risks, failures, and handoffs through a routed one-question-at-a-time grilling workflow. Use when the user asks to grill, challenge, harden, clarify, pre-review, premortem, design test cases, prepare implementation, diagnose why a plan failed, or decide whether work is ready for an agent.

zengxiansheng007 Bundle

Build and sharpen a project's domain model. Use when the user wants to pin down domain terminology or a ubiquitous language, record an architectural decision, or when another skill needs to maintain the domain model.

zengxiansheng007 Bundle

Convert ui-test-packet into a stable A2A-compatible handoff_packet, prompt_packet, and artifact_refs bundle for Codex to keep refining Midscene requests. Use when the user asks for a UI-test交接协议层, prompt转换层, packet to packet mapping, failure_history preservation, retry_history preservation, or artifact reference hygiene. Do not use for real A2A transport, browser execution, or checkpoint runtime.

wolski Bundle

Runs a panel-style multi-agent code review. Dispatches specialists for GoF patterns, antipatterns and declarative boundaries, function complexity, mixed abstraction and public API contracts, and architecture boundaries the repository declares about itself; then synthesizes a balanced, prioritized report. Supports branch diffs, specific paths, whole codebases, and commit ranges. Use when the user asks for a multi-agent, panel, deep, GoF, antipattern, API-design, architecture-boundary, layering, whole-codebase, full, or multi-perspective code review, or invokes /multiagent-review.

dzhokhov Skill

Research – Skill of Quality LLM Research

dzhokhov Bundle

Creates a new vault project and the full starter documentation set: plan.md, tasks.md, context.md, log.md, README.md, plus index updates. The agent writes the plan itself using the autonomous LLM-agent planning methodology: Intent Lock, Owner Interaction Policy, milestones with acceptance conditions, Quality Criteria, and Drift Guard. Use when the owner says: "create a project", "start a project", "turn this into a project", "capture this as a project", "set up a project for X", or equivalent wording. Works in two scenarios: the project is extracted from the current conversation, or the owner asks for a new project from scratch. Do not use for meeting processing, parking/resume flows, or ordinary continuation of an existing project.

anubisrooster Bundle

Run long-lived or unattended agent jobs (digest publishers, crawlers, report builders, watchers) inside a Herdr pane so they survive terminal disconnects, lid close, and reboots, and stay observable (working/blocked/done). Use when a job runs longer than a few minutes, is launched by a scheduler, or must be inspected after the fact. Requires the Herdr CLI (herdr.dev).

anubisrooster Skill

Create Jira Epics/Features and child sprint issues via the mcp-atlassian server (self-hosted Jira Data Center), enforcing a mandatory field set (Summary, Description, Epic Link/Parent Link, Estimate, Components, Labels, Function area, Business Value, Acceptance Criteria, Definition of Ready, Definition of Done) and a hierarchical WBS numbering rule (Initiative X.0 -> Epic/Feature X.Y -> Story/Task/Test/Defect X.Y.Z) regardless of what Jira's screens require. Enforces the three-orthogonal-dimensions model (Pillars=Components, Teams=Labels, Programs=Initiatives), 8 pillar-based components, team/product/BU labels, cross-pillar work rules, and initiative gating. Always shows the proposed payload to the user for approval before creating. Use when the user asks to create, draft, file, scaffold, or open a Jira Epic, Feature, Story, Task, Test, Defect, or sub-task under an Epic/Feature/Initiative.

anubisrooster Skill

Extend the mcp-atlassian MCP server to connect Claude/Cursor to a self-hosted Confluence Data Center instance. Use this skill whenever the user mentions Confluence Data Center, self-hosted Confluence, on-prem Confluence, Confluence behind a VPN, setting up Confluence MCP, Confluence personal access tokens, searching Confluence wiki pages, creating or updating Confluence pages from Claude, CQL queries, Confluence spaces, or any workflow that involves reading or writing documentation in a self-hosted Confluence. Also trigger when the user asks to extend their existing mcp-atlassian Jira setup with Confluence, or wants to connect Claude/Cursor to an internal wiki. Complements the jira-create-issues skill — both share the same mcp-atlassian MCP server process.

anubisrooster Bundle

Daily LATC Confluence digest for Mike Fink's personal space (~mfink). Runs as a Windows Scheduled Task weekdays at 8:00 AM Eastern. Uses the Cursor SDK (model grok-4.6) to scrape LATC Confluence updates, score/cluster them, join Jira where possible, publish a dated digest under the LATC Daily Digest hub in Mike Fink's writing-voice (formal Confluence), and point the hub Latest link. Use when setting up, troubleshooting, or changing digest format, cadence, filters, voice, or schedule.

starchild-ai-agent Skill

Director-level Seedance 2 video prompt maker for 15-second cinematic multi-shot CGI animation prompts.

starchild-ai-agent Skill

FVG-Delta Crypto Signal Engine — Agent / Maintainer Guide (v5.5)

stacksjs Skill

Use when working with the event system in a Stacks application — dispatching events, listening for events, model events, wildcard listeners, the event emitter, or event-driven architecture. Covers @stacksjs/events, app/Events.ts, app/Listener.ts, and app/Listeners/.

stacksjs Skill

Use when implementing validation in Stacks — type guards (isString, isNumber, isBoolean, isObject, isArray, isFunction, etc.), numeric checks (isPositive, isEven, isInteger), the schema builder for model attribute validation, or request validation. Covers @stacksjs/validation.

taracodlabs Bundle

Generate images using Stable Diffusion via HuggingFace Diffusers library locally or via API

changhochien Skill

Merger Model

foryourhealth111-pixel Bundle

Production-ready reinforcement learning algorithms (PPO, SAC, DQN, TD3, DDPG, A2C) with scikit-learn-like API. Use for standard RL experiments, quick prototyping, and well-documented algorithm implementations. Best for single-agent RL with Gymnasium environments. For high-performance parallel training, multi-agent systems, or custom vectorized environments, use pufferlib instead.

foryourhealth111-pixel Bundle

Structured hypothesis formulation from observations. Use when you have experimental observations or data and need to formulate testable hypotheses with predictions, propose mechanisms, and design experiments to test them. Also owns explicit HypoGeniC-style or automated LLM-driven hypothesis generation/testing requests inside this single skill. For open-ended ideation use scientific-brainstorming.

arch1esun Skill

Use when starting a new round (substrate-touching / fix-round / admin) to generate a complete handoff doc from scope + prior-round context. Dispatches the planner sub-agent and validates output before writing.

arch1esun Bundle

Runs the mechanical quality gates that the arcgentic state machine requires for state transitions. Invoked indirectly by transition.sh OR directly by orchestrator agent before declaring a state transition. Use when about to call transition.sh OR when manually verifying that a round artifact meets the gate criteria. Each gate is a single Bash script; output is PASS/FAIL with reason.

arch1esun Skill

Use when a round references agency-agents catalogs, role-family routing, multi-agent identity prompts, or English/Chinese specialist role catalogs.

arch1esun Bundle

Mandatory prelude for every role at the start of a round. Inventories available skills, MCP servers, plugins, sub-agents, and external references. Outputs which tools will be used in this round and which were considered but rejected, with reasons. Use as the very first action of every role skill in an arcgentic-managed project.

evanoman Skill

Code review using a sub-agent with structured rubric. Use when the user wants a code review of current changes or a specific file.

evanoman Skill

Generate a structured handoff note for project continuity. Use when ending a session and wanting to capture context for the next agent or session.

evanoman Bundle

Orchestrate persistent, cross-harness subagents through a shared local task broker with grouped fan-out, event-driven waiting, bounded result collection, continuation, interruption, recovery, and usage evidence. Use when the user explicitly asks for Overmind v2, coordinated parallel work, a Claude-to-Codex or Codex-to-Claude worker, background agents that must survive the parent session, a subagent bake-off, or reliable status tracking without model-driven polling.

evanoman Bundle

Run parallelized multi-agent research on any topic via a deterministic workflow. Use when the user asks to research a technology, strategy, architecture, library, approach, or any complex question that benefits from partitioning the search space across multiple sub-agents. Each agent writes its own report, then a synthesis report distills the high-value findings.