| name | toucan-mcp-tool-agentic-dataset |
| title | TOUCAN: Synthesizing Tool-Agentic Data from MCP Environments |
| version | 0.0.2 |
| engine | skillxiv-v0.0.2-claude-opus-4.6 |
| license | MIT |
| url | https://arxiv.org/abs/2510.01179 |
| keywords | ["MCP","agent-training","dataset-synthesis","tool-use","LLM-agents"] |
| description | Generate 1.5M realistic agent training trajectories from 495 real-world MCP servers without human annotation. Use to build tool-agentic training datasets with authentic tool execution, multi-turn conversations, and error handling. |
TOUCAN: Synthesizing Tool-Agentic Data from MCP Environments
TOUCAN introduces the largest open-source tool-agentic dataset with 1.5M trajectories generated from real-world MCP (Model Context Protocol) servers. By leveraging authentic MCP execution rather than simulation, the dataset enables LLM agent training with realistic tool behavior, error patterns, and complex workflows.
Core Architecture
- Real MCP execution: 495 servers with actual tool behavior (no simulation)
- 1.5M trajectories: Multi-turn conversations with parallel tool calls
- Quality filtering: Six-dimension validation (relevance, correctness, completeness, diversity, executability, safety)
- Diverse interactions: Error recovery, irrelevant queries, multi-step reasoning
- State-of-the-art results: Fine-tuned 32B models outperform GPT-4.5-Preview on multi-turn benchmarks
Implementation Steps
Setup MCP server discovery and trajectory synthesis:
from toucan import MCPDatasetSynthesizer, QualityFilter
synthesizer = MCPDatasetSynthesizer(
num_servers=495,
trajectory_budget=1_500_000,
parallel_workers=128,
server_timeout=30
)
servers = synthesizer.discover_servers(
exclude_api_key_dependent=True,
max_per_category=50
)
quality_filter = QualityFilter(
dimensions={
"relevance": 0.85,
"correctness": 0.90,
"completeness": 0.80,
"diversity": 0.75,
"executability": 0.95,
"safety": 1.0
}
)
Generate trajectories with error handling and quality assurance:
