multi-agent-ralph-loop

Full orchestration workflow with swarm mode: evaluate -> clarify -> classify -> persist -> plan mode -> spawn teammates -> execute -> validate -> retrospective. Use when: (1) implementing features, (2) complex refactoring, (3) multi-file changes, (4) tasks requiring coordination. Triggers: /orchestrator, /orch, 'orchestrate', 'full workflow', 'implement feature'.

Comprehensive research skill using Zai MCP web search and native Claude Code tools

Smart Forking - Find and fork from relevant historical sessions using parallel memory search across vault, handoffs, and ledgers

Apply adversarial opposite-analysis to plans, specs, architecture, code changes, and claims. Use when the user asks for adversarial review, opposing analysis, contrarian review, red-team reasoning, or Z.ai and MiniMax cross-checks through the Ralph MCP router.

Test case mutation and variation generator for adversarial testing

Patterns for using Context7 MCP for library documentation (v2.25)

Codebase defense analysis system for security profiling

Google Gemini CLI orchestration (v0.22.0+) for AI-assisted development. Capabilities: second opinion/cross-validation, real-time Google Search grounding, codebase architecture analysis with codebase_investigator, Gemini 3 model access, extensions support (Conductor, Endor Labs), parallel code generation, code review from different perspective. INTEGRATED WITH TASK PRIMITIVE - creates traceable tasks in claude-task-viewer. Actions: query, search, analyze, generate, review with Gemini. Keywords: Gemini CLI, Gemini 3, google_web_search, codebase_investigator, second opinion, cross-validation, web research, current information, parallel AI, code review, architecture analysis, gemini prompt, AI comparison, real-time search, alternative perspective, extensions, Conductor. Use when: needing second AI opinion, searching current web information, analyzing codebase architecture, generating code in parallel, getting alternative code review, researching current events/docs, using Gemini extensions.

[DEPRECATED] Use /research skill with Zai MCP instead - Optimal patterns for MiniMax MCP tools

Tree of Attacks with Pruning for systematic code analysis

Visualize task dependencies and progress (Gastown-style)

React and Next.js performance optimization guidelines from Vercel Engineering. Use when writing, reviewing, or refactoring React/Next.js code. Triggers on tasks involving React components, Next.js pages, data fetching, bundle optimization, or performance improvements.

Manage git worktrees with PR workflow and multi-agent review (Claude + Codex). Use when developing features in isolation with easy rollback.

Autonomous experiment loop: modifies code, runs experiments, evaluates metrics, keeps improvements. Inspired by karpathy/autoresearch + pi-autoresearch + autoexp. Triggers: /autoresearch, 'auto research', 'optimize continuously', 'experiment loop', 'autonomous optimization'.

Create technical and product diagrams — architecture, flowchart, sequence, state machine, ER / data model, timeline, swimlane, quadrant, nested, tree, layer stack, venn, pyramid — as standalone HTML files with inline SVG. Ships with a neutral editorial skin and a first-run gate that prompts users to customize the style guide (colors, fonts) from their own website before generating. Includes annotation-callout primitive and optional sketchy variant.

Create professional, dark-themed architecture diagrams as standalone HTML files with SVG graphics. Use when the user asks for system architecture diagrams, infrastructure diagrams, cloud architecture visualizations, security diagrams, network topology diagrams, or any technical diagram showing system components and their relationships.

Classifies task complexity (1-10) for model and agent routing

Bug hunting with Codex CLI Use when: (1) /bugs is invoked, (2) task relates to bugs functionality.

Determines WHAT context an agent needs and packages it optimally. Actions: analyze (identify needed context), load (assemble from sources), prune (trim to token budget), inject (write to .claude/context-payload.md). Use when: (1) before spawning teammates, (2) context window is limited, (3) multi-source context assembly. Triggers: /context-engineer, 'prepare context', 'package context', 'context for agent'.

Learn patterns from a specific GitHub repository. Clones, analyzes code structure, extracts patterns, populates procedural memory AND syncs to Obsidian vault for Graph View visualization. Use for: targeted learning from known quality repos, quick knowledge acquisition, specific pattern extraction. Triggers: /repo-learn, /curator-repo-learn, 'learn from repo'.

Full curator pipeline for autonomous learning from quality repositories. Executes: discovery → scoring → ranking → ingest → learn → vault sync. Writes to procedural memory AND Obsidian vault for Graph View visualization and graduation pipeline. Use for: populating procedural memory with domain patterns, first-time domain learning, comprehensive knowledge building. Triggers: /curator full, 'learn patterns from repos', 'build knowledge base'.

Eval-Driven Development (EDD) Framework v2.87.0 - Define-before-implement pattern with structured evals. Provides workflow: Define specifications → Implement features → Verify against evals. Components: TEMPLATE.md for eval definitions, edd.sh CLI script, /edd skill invocation. Check types: CC- (Capability), BC- (Behavior), NFC- (Non-Functional). Integrates with orchestrator workflow for quality-first development. Keywords: evals, define, implement, verify, capability checks, behavior checks, non-functional checks, template, quality assurance, test-driven, specification. Use when: defining new features with structured evals, implementing with verification requirements, creating quality specifications, TDD-style workflow with evals.

Custom skill for glm-mcp

Custom skill for kaizen

Custom skill for minimax

Autonomous batch task execution with PRD parsing, task decomposition, and continuous execution until all tasks complete. Uses /orchestrator internally. Stops only for major failures (no internet, token limit, system crash). Use when: (1) processing task lists autonomously, (2) PRD-driven development, (3) batch feature implementation. Triggers: /task-batch, 'batch tasks', 'process PRD', 'run task queue'.

Custom skill for testing-anti-patterns

Ask clarifying questions when requirements are underspecified

Generate usage report for MiniMax and token optimization Use when: (1) /audit is invoked, (2) task relates to audit functionality.

Session checkpoint management: save, restore, list, clear state snapshots

Remove AI-generated code slop from a branch. Use when cleaning up AI-generated code, removing unnecessary comments, defensive checks, or type casts. Checks diff against main and fixes style inconsistencies.

OpenAI Codex CLI orchestration for AI-assisted development using gpt-5.3-codex model family. Model variants: gpt-5.3-codex (medium), gpt-5.3-codex-high, gpt-5.3-codex-xhigh. Capabilities: code generation, refactoring, automated editing, parallel task execution, session management, code review, architecture analysis, and MCP integration. Actions: analyze, implement, review, fix, refactor with Codex. Keywords: Codex CLI, gpt-5.3-codex, codex exec, code generation, refactoring, parallel execution, session resume, code review, second opinion, independent review, architecture validation, Context7 MCP. Use when: delegating complex code tasks to Codex, running multi-agent workflows, executing automated reviews, implementing features with AI assistance, resuming previous sessions, querying OpenAI documentation. Triggers: 'use codex', 'codex exec', 'run with codex', 'codex resume', 'implement with codex', 'review with codex', 'codex docs'.

Use when writing or improving README files. Not all READMEs are the same — provides templates and guidance matched to your audience and project type.

Code quality analyzer that identifies "slop" - code violating established coding principles - and suggests concrete improvements. Analyzes code against principles like KISS, YAGNI, SOLID, DRY, and provides before/after fix examples.

Model-agnostic parallel execution with Agent Teams coordination

GLM-5 Agent Teams skill for spawning teammates with thinking mode

Access OpenAI developer documentation via Context7 MCP. Provides up-to-date docs for Codex CLI, OpenAI API, Python/Node SDKs, Agents SDK, and MCP configuration. Use when: (1) configuring Codex CLI or MCP servers, (2) writing OpenAI API integrations, (3) building agents with OpenAI SDKs, (4) troubleshooting Codex execution. Triggers: 'openai docs', 'codex documentation', 'openai api reference', 'codex mcp', 'agents sdk'.

Product Requirements Document generation and management with INVEST-compliant user stories

When the user wants to create or update a README.md file for a project. Also use when the user says "write readme," "create readme," "document this project," "project documentation," or asks for help with README.md. This skill creates absurdly thorough documentation covering local setup, architecture, and deployment.

Senior Blockchain Architect research agent using Zai MCP for comprehensive analysis of EVM chains, perpetual DEX architectures, CEX integrations, and DeFi-TradFi bridges. Use for: blockchain research, protocol comparisons, technical feasibility studies, security audits, compliance analysis, architecture blueprints. Triggers: /research-blockchain, 'blockchain research', protocol comparisons.