12fdd18223
Deleted: - DQN::compute_loss_internal (280 lines) — old Candle forward+loss - DQN::train_step (55 lines) — old Candle training step - DQN::compute_gradients (47 lines) — old gradient accumulation - ComputeLossResult struct — only used by deleted functions - RegimeConditionalDQN::train_step (65 lines) — old dispatch - RegimeConditionalDQN::train_step_gpu_regime (100 lines) — old GPU path - RegimeConditionalDQN::compute_gradients_gpu (130 lines) — old regime gradients - RegimeConditionalDQN::compute_gradients (92 lines) — old dispatch - DQNAgentType::train_step dispatch — dead - DQNAgentType::compute_gradients dispatch — dead - GpuDqnTrainer::upload_batch (71 lines) — old CPU→GPU upload - train_step.rs (500 lines) — entire module including ensure_fused_ctx - dqn_benchmark.rs — used old train_step - examples.rs — used old train_step - validation/adapters.rs (289 lines) — used old train_step - dqn/trainable_adapter.rs — used old train_step - gpu_smoketest.rs — tested old train_step - Gradient accumulation path in training_loop.rs (144 lines) - IQN d_h_s2().clone() → raw pointer (zero alloc) - Causal intervention format! string alloc removed - Dead HER relabel functions (320 lines) Kept: - ensure_fused_ctx logic inlined into training_loop.rs - set_noise_sigma_scale re-added to RegimeConditionalDQN Fixed: - GpuReplayBuffer max_batch_size wired from batch_size parameter (was hardcoded 1024, blocking batch_size=8192) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
3.9 KiB
3.9 KiB
SPARC Modes Overview
SPARC (Specification, Planning, Architecture, Review, Code) is a comprehensive development methodology with 17 specialized modes, all integrated with MCP tools for enhanced coordination and execution.
Available Modes
Core Orchestration Modes
- orchestrator: Multi-agent task orchestration
- swarm-coordinator: Specialized swarm management
- workflow-manager: Process automation
- batch-executor: Parallel task execution
Development Modes
- coder: Autonomous code generation
- architect: System design
- reviewer: Code review
- tdd: Test-driven development
Analysis and Research Modes
- researcher: Deep research capabilities
- analyzer: Code and data analysis
- optimizer: Performance optimization
Creative and Support Modes
- designer: UI/UX design
- innovator: Creative problem solving
- documenter: Documentation generation
- debugger: Systematic debugging
- tester: Comprehensive testing
- memory-manager: Knowledge management
Usage
Option 1: Using MCP Tools (Preferred in Claude Code)
// Execute SPARC mode directly
mcp__claude-flow__sparc_mode {
mode: "<mode>",
task_description: "<task>",
options: {
// mode-specific options
}
}
// Initialize swarm for advanced coordination
mcp__claude-flow__swarm_init {
topology: "hierarchical",
strategy: "auto",
maxAgents: 8
}
// Spawn specialized agents
mcp__claude-flow__agent_spawn {
type: "<agent-type>",
capabilities: ["<capability1>", "<capability2>"]
}
// Monitor execution
mcp__claude-flow__swarm_monitor {
swarmId: "current",
interval: 5000
}
Option 2: Using NPX CLI (Fallback when MCP not available)
# Use when running from terminal or MCP tools unavailable
npx claude-flow sparc run <mode> "task description"
# For alpha features
npx claude-flow@alpha sparc run <mode> "task description"
# List all modes
npx claude-flow sparc modes
# Get help for a mode
npx claude-flow sparc help <mode>
# Run with options
npx claude-flow sparc run <mode> "task" --parallel --monitor
Option 3: Local Installation
# If claude-flow is installed locally
./claude-flow sparc run <mode> "task description"
Common Workflows
Full Development Cycle
Using MCP Tools (Preferred)
// 1. Initialize development swarm
mcp__claude-flow__swarm_init {
topology: "hierarchical",
maxAgents: 12
}
// 2. Architecture design
mcp__claude-flow__sparc_mode {
mode: "architect",
task_description: "design microservices"
}
// 3. Implementation
mcp__claude-flow__sparc_mode {
mode: "coder",
task_description: "implement services"
}
// 4. Testing
mcp__claude-flow__sparc_mode {
mode: "tdd",
task_description: "test all services"
}
// 5. Review
mcp__claude-flow__sparc_mode {
mode: "reviewer",
task_description: "review implementation"
}
Using NPX CLI (Fallback)
# 1. Architecture design
npx claude-flow sparc run architect "design microservices"
# 2. Implementation
npx claude-flow sparc run coder "implement services"
# 3. Testing
npx claude-flow sparc run tdd "test all services"
# 4. Review
npx claude-flow sparc run reviewer "review implementation"
Research and Innovation
Using MCP Tools (Preferred)
// 1. Research phase
mcp__claude-flow__sparc_mode {
mode: "researcher",
task_description: "research best practices"
}
// 2. Innovation
mcp__claude-flow__sparc_mode {
mode: "innovator",
task_description: "propose novel solutions"
}
// 3. Documentation
mcp__claude-flow__sparc_mode {
mode: "documenter",
task_description: "document findings"
}
Using NPX CLI (Fallback)
# 1. Research phase
npx claude-flow sparc run researcher "research best practices"
# 2. Innovation
npx claude-flow sparc run innovator "propose novel solutions"
# 3. Documentation
npx claude-flow sparc run documenter "document findings"