npx skills add ruvnet/ruflo --skill agent-agent --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
Instructs an AI agent to perform goal-oriented action planning (GOAP) and multi-agent coordination using a sublinear optimization framework. It covers dynamic goal decomposition, graph-based action modeling, goal prioritization (PageRank), temporal advantage planning, and A* search augmented with sublinear heuristics. It includes multi-agent coordination workflows (swarm initialization, agent spawning, task orchestration) and consensus-based decision making. It also provides advanced planning workflows such as hierarchical decomposition and dynamic replanning, plus learning from execution and gaming AI-style behavior trees for GOAP planning.
How it works
- State space is modeled with a WorldState and a set of Actions, each with costs, preconditions, and effects.
- Action Graph Construction builds an adjacency matrix reflecting possible transitions between actions, then analyzes it with a solver (mcp__sublinear_time_solver__analyzeMatrix).
- Goal prioritization uses mcp__sublinear_time_solver__pageRank on the action graph to rank goals and actions by importance.
- Temporal advantage planning invokes mcp__sublinear_time_solver__predictWithTemporalAdvantage and mcp__sublinear_time_solver__validateTemporalAdvantage to forecast feasible solutions.
- A* search is enhanced by a sublinear heuristic, leveraging optimizedHeuristic and a sublinear solver during planning.
- Multi-Agent Coordination includes swarm initialization via mcp__claude_flow__swarm_init, agent_spawn calls for coordinator/analyst/optimizer, and task orchestration through mcp__claude_flow__task_orchestrate; consensus-based decision making uses a Neumann-method solver on a consensus matrix.
- Advanced workflows include hierarchical goal decomposition using sandboxed goal decomposition, dependency matrices, and a PageRank-based execution order; dynamic replanning is implemented with a DynamicPlanner class and memory usage integration for goap-patterns.
- Learning from execution provides methods to assess plan effectiveness, store successful patterns, and train a neural model via mcp__flow_nexus__neural_train; it can retrieve similar patterns from memory.
- Gaming AI integration introduces a GOAPBehaviorTree with behavior-tree nodes for plan generation and execution.
When to use it
- When you need a GOAP-based planner with dynamic goal decomposition, prioritization, and real-time replanning capabilities.
- When coordinating multiple agents to achieve a complex goal using a coordinated planning workflow.
- When you want to combine sublinear optimization techniques with classic planning search (A*/heuristics) for efficient path to deployment readiness.
What it can touch
- Tools and functions named in the skill, including:
- mcp__sublinear_time_solver__solve
- mcp__sublinear_time_solver__pageRank
- mcp__sublinear_time_solver__analyzeMatrix
- mcp__sublinear_time_solver__predictWithTemporalAdvantage
- mcp__sublinear_time_solver__validateTemporalAdvantage
- mcp__sublinear_time_solver__estimateEntry
- mcp__sublinear_time_solver__calculateLightTravel
- mcp__sublinear_time_solver__demonstrateTemporalLead
- mcp__flow_nexus__sandbox_create
- mcp__flow_nexus__neural_train
- mcp__flow_nexus__memory_usage
- mcp__flow_nexus__memory_search
- mcp__claude_flow__swarm_init
- mcp__claude_flow__agent_spawn
- mcp__claude_flow__workflow_create
- mcp__claude_flow__task_orchestrate
- mcp__claude_flow__agent_spawn
- mcp__flow_nexus__neural_train
- mcp__sublinear_time_solver__analyzeMatrix
- mcp__sublinear_time_solver__solve
- mcp__sublinear_time_solver__pageRank
- mcp__sublinear_time_solver__predictWithTemporalAdvantage
- mcp__sublinear_time_solver__validateTemporalAdvantage
- mcp__sublinear_time_solver__estimateEntry
- mcp__sublinear_time_solver__calculateLightTravel
- mcp__sublinear_time_solver__demonstrateTemporalLead These are the concrete function/tool names the skill references.
Caveats
- License: MIT
- Declared tools: claude-code, codex
- The skill description and workflow rely on a set of interdependent sublinear solver tools and Claude Flow integration, potentially requiring compatible runtimes to execute as described.
--- name: sublinear-goal-planner description: "Goal-Oriented Action Planning (GOAP) specialist that dynamically creates intelligent plans to achieve complex objectives. Uses gaming AI techniques to discover novel solutions by combining actions in creative ways. Excels at adaptive replanning, multi-step reasoning, and finding optimal paths through complex state spaces." color: cyan --- A sophisticated Goal-Oriented Action Planning (GOAP) specialist that dynamically creates intelligent plans to achieve complex objectives using advanced graph analysis and sublinear optimization techniques. This agent transforms high-level goals into executable action sequences through mathematical optimization, temporal advantage prediction, and multi-agent coordination. ## Core Capabilities ### 🧠 Dynamic Goal Decomposition - Hierarchical goal breakdown using dependency analysis - Graph-based representation of goal-action relationships - Automatic identification of prerequisite conditions and dependencies - Context-aware goal prioritization and sequencing ### ⚡ Sublinear Optimization - Action-state graph optimization using advanced matrix operations - Cost-benefit analysis through diagonally dominant
- Core Capabilities
- 🧠 Dynamic Goal Decomposition
- ⚡ Sublinear Optimization
- 🎯 Intelligent Prioritization
- 🔮 Predictive Planning
- 🤝 Multi-Agent Coordination
- Primary Tools
- Sublinear-Time Solver Tools
- Claude Flow Integration Tools
- Workflow
- 1. State Space Modeling
- 2. Action Graph Construction
- 3. Goal Prioritization with PageRank
- 4. Temporal Advantage Planning
What does the agent-agent skill do?
Agent skill for agent - invoke with $agent-agent
How do I install it?
Run `npx skills add ruvnet/ruflo --skill agent-agent --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From ruvnet/ruflo, a repository with 67,015 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.