Intelligent
Sub-Agent Orchestration
A DAG-based orchestration system that decomposes complex requests into parallel and sequential tasks, executed by 13 specialized sub-agents with model routing and cost optimization.
DAG-Based Task Decomposition
Complex requests are automatically broken into a directed acyclic graph (DAG) of tasks — independent tasks run in parallel, dependent tasks wait for their predecessors.
13 Purpose-Built Sub-Agents
Each sub-agent is a domain expert with its own model assignment, MCP tool access, and system prompt — assembled dynamically by the orchestrator.
Data Analyst
Analyzes datasets with pandas — statistics, correlations, rankings, and charts. Generates matplotlib visualizations.
Python ExecutorResearcher
Web search and content fetching with structured summaries and citations. Cross-references multiple sources.
Web SearchMarket Researcher
Economic conditions, market trends, competitor activity analysis for specific geographic regions and industries.
Web SearchCompetitive Analyst
Competitive landscape analysis — market share, feature comparison, pricing models, and SWOT analysis.
ResearchFinancial Modeler
Financial models, TAM/SAM/SOM analysis, pricing optimization, and revenue projections with Excel outputs.
Python + ExcelRegulatory Researcher
Regulatory, legal, and compliance requirements analysis for market entry across jurisdictions.
ResearchWriter
Documents, reports, and articles — Word docs, markdown, and structured reports with professional formatting.
FilesystemReport Writer
Executive PowerPoint presentations designed for board-level audiences with data visualization.
PPTX SkillCompliance Writer
Audit-ready compliance reports mapped to SOC 2, OWASP, PCI-DSS frameworks with control matrices.
FilesystemCode Reviewer
Comprehensive code review for bugs, security vulnerabilities, code style, and best practices.
Code AnalysisCode Scanner
Security scanning — vulnerability detection, hardcoded secrets, injection risks, and dependency analysis.
SecurityRemediation Coder
Generates secure code patches in diff format to fix identified vulnerabilities with detailed explanations.
Code GenEvaluator
Validates sub-agent output quality against the original request for completeness, accuracy, and relevance.
Quality GateIntelligent Model Routing
Each sub-agent is automatically assigned the optimal model based on task complexity, cost tier, and remaining budget.
Haiku
Claude Haiku 4.5
- Research & web search
- File discovery
- Linting & validation
- Simple lookups
Sonnet
Claude Sonnet 4.6
- Code review & generation
- Data analysis
- Report writing
- Complex reasoning
Opus
Claude Opus 4.6
- Orchestration & planning
- Architecture decisions
- Complex decomposition
- Multi-agent coordination
Create Agents in Minutes
Define sub-agents with a simple YAML file — model selection, tool access, skills, and behavioral constraints. Drop in the file and restart.
- YAML-based declarative configuration
- REST API for runtime agent creation
- Natural language agent generation
- Per-agent MCP tool sandboxing
- Shared or isolated workspace modes
- Configurable iteration limits & timeouts
name: "data-analyst"
description: "Analyzes datasets with pandas"
model: "sonnet"
temperature: 0.3
max_tokens: 4096
system_prompt: |
You are an expert data analyst.
Load, analyze, and visualize data.
mcp_servers:
- "python-executor"
- "filesystem"
skills:
- "xlsx"
max_iterations: 15
timeout_seconds: 600
workspace_mode: "shared"
Orchestrate Complex Workflows
Automatically
Let the DAG-based orchestrator decompose, parallelize, execute, and validate — while you focus on the results.
Learn About the 5-Stage Process →