task — Claude Code command: orchestrate complex workflows
task — a slash command for multi-stage project execution with seven expert personas, intelligent MCP server routing, and cross-session context preservation. User-invoked, runs to completion, then stops.
MIT ★ 23,950 updated 2026-09-27
.claude/commands/task.mdWhat task does
The task command parses requirements, activates seven expert roles (architect, analyzer, frontend, backend, security, devops, project manager), and routes them to appropriate MCP servers — Sequential for multi-step analysis, Context7 for framework patterns, Magic for UI/UX coordination, Playwright for test automation, Morphllm for large-scale transformations, Serena for memory across sessions.
Execution follows five stages: analyze task scope, delegate to competencies, coordinate with optional parallelism, validate against quality gates, optimize based on outcomes. Three strategies available: systematic for comprehensive coverage, agile for iterative delivery, enterprise for governance compliance. Unlike /sc:pm (background session-level orchestration), task is user-initiated with explicit start and end — user invokes it, the command executes, returns a completion report, and stops.
If an MCP server is unavailable, graceful degradation applies: native reasoning replaces Sequential, web-based docs lookup replaces Context7, built-in tools handle the rest. All subtasks tracked through TodoWrite with Epic → Story → Task → Subtask hierarchy.
What it helps with
- break a large project into coordinated milestones and assign them to different specialists
- execute work that spans multiple domains (security, performance, testing, design) in one flow
- retain task context and progress across separate work sessions
- run independent subtasks in parallel and merge the results
- choose the right execution strategy from rapid prototype to full production with compliance
Example prompts for the agent
- /sc:task create “enterprise authentication platform” --strategy systematic --parallel
- /sc:task execute “frontend refactor” --strategy agile --delegate
- /sc:task execute “payment gateway integration” --parallel
How to install task
Manually: put the files in place
.claude/commands/task.mdCall it with /task. For all projects use ~/.claude/commands/
Manually: put the files in place
~/.codex/prompts/task.mdCall it with /prompts:task. Codex marks this format deprecated in favour of skills
Manually: put the files in place
.github/prompts/task.prompt.mdManually: put the files in place
.gemini/commands/task.tomlGemini stores commands as TOML: the prompt text goes into the prompt field
Questions about task
How do I install task in Claude Code?
Place the task.md file in your project's .claude/commands/ directory. Claude Code will recognize it and make the command available as /sc:task.
What is the difference between task and pm?
/sc:pm runs at session level as continuous background orchestration. /sc:task is a discrete user-invoked execution: it starts when you call it, completes the goal, and stops, ready for your next instruction.
Do I need all the MCP servers for task to work?
Sequential, Context7, Magic, Playwright, Morphllm, and Serena are preferred but optional. If any is missing, the command degrades gracefully: built-in reasoning replaces Sequential, documentation lookup replaces Context7, native tools handle the rest.
Source text
task.md · MIT
--- name: task description: "Execute complex tasks with intelligent workflow management and delegation" category: special complexity: advanced mcp-servers: [sequential, context7, magic, playwright, morphllm, serena] personas: [architect, analyzer, frontend, backend, security, devops, project-manager] --- # /sc:task - Enhanced Task Management ## Triggers - Complex tasks requiring multi-agent coordination and delegation - Projects needing structured workflow management and cross-session persistence - Operations requiring intelligent MCP server routing and domain expertise - Tasks benefiting from systematic execution and progressive enhancement ## Usage ``` /sc:task [action] [target] [--strategy systematic|agile|enterprise] [--parallel] [--delegate] ``` ## Behavioral Flow 1. **Analyze**: Parse task requirements and determine optimal execution strategy 2. **Delegate**: Route to appropriate MCP servers and activate relevant personas 3. **Coordinate**: Execute tasks with intelligent workflow management and parallel processing 4. **Validate**: Apply quality gates and comprehensive task completion verification 5. **Optimize**: Analyze performance and provide enhancement recommendations Key behaviors: - Multi-persona coordination across architect, frontend, backend, security, devops domains - Intelligent MCP server routing (Sequential, Context7, Magic, Playwright, Morphllm, Serena) - Systematic execution with progressive task enhancement and cross-session persistence - Advanced task delegation with hierarchical breakdown and dependency management ## MCP Integration - **Sequential MCP**: Complex multi-step task analysis and systematic execution planning - **Context7 MCP**: Framework-specific patterns and implementation best practices - **Magic MCP**: UI/UX task coordination and design system integration - **Playwright MCP**: Testing workflow integration and validation automation - **Morphllm MCP**: Large-scale task transformation and pattern-based optimization - **Serena MCP**: Cross-session task persistence and project memory management - **Availability fallback**: The MCP servers named here are preferred tooling, not hard requirements — when one is not installed, degrade gracefully (native reasoning in place of Sequential, official docs via web lookup in place of Context7, native tools otherwise) instead of routing work to absent servers. ## Tool Coordination - **TodoWrite**: Hierarchical task breakdown and progress tracking across Epic → Story → Task levels - **Task**: Advanced delegation for complex multi-agent coordination and sub-task management - **Read/Write/Edit**: Task documentation and implementation coordination - **sequentialthinking**: Structured reasoning for complex task dependency analysis ## Key Patterns - **Task Hierarchy**: Epic-level objectives → Story coordination → Task execution → Subtask granularity - **Strategy Selection**: Systematic (comprehensive) → Agile (iterative) → Enterprise (governance) - **Multi-Agent Coordination**: Persona activation → MCP routing → parallel execution → result integration - **Cross-Session Management**: Task persistence → context continuity → progressive enhancement ## Examples ### Complex Feature Development ``` /sc:task create "enterprise authentication system" --strategy systematic --parallel # Comprehensive task breakdown with multi-domain coordination # Activates architect, security, backend, frontend personas ``` ### Agile Sprint Coordination ``` /sc:task execute "feature backlog" --strategy agile --delegate # Iterative task execution with intelligent delegation # Cross-session persistence for sprint continuity ``` ### Multi-Domain Integration ``` /sc:task execute "microservices platform" --strategy enterprise --parallel # Enterprise-scale coordination with compliance validation # Parallel execution across multiple technical domains ``` ## Boundaries **Will:** - Execute complex tasks with multi-agent coordination and intelligent delegation - Provide hierarchical task breakdown with cross-session persistence - Coordinate multiple MCP servers and personas for optimal task outcomes **Will Not:** - Execute simple tasks that don't require advanced orchestration - Compromise quality standards for speed or convenience - Operate without proper validation and quality gates ## CRITICAL BOUNDARIES **USER-INVOKED DISCRETE TASK EXECUTION** This command executes specific tasks when explicitly invoked by user. **Difference from /sc:pm**: - `/sc:pm` = session-level orchestration (background monitoring, continuous) - `/sc:task` = user-invoked discrete execution (explicit start/end) **Behavior**: - User invokes `/sc:task [description]` - Execute the specific task using multi-agent coordination - **STOP when task is complete** - do not continue to next tasks without user input **Completion Criteria**: - Task objective achieved - All sub-tasks marked completed in TodoWrite - Validation passed **Output**: Task completion report with: - What was accomplished - Files modified - Tests status (if applicable) **Next Step**: User decides next action. May invoke another `/sc:task` or use specific commands.
© SuperClaude-Org · MIT · SuperClaude-Org/SuperClaude_Framework