dingo saas

Managing SaaS datasets and LLM evaluation experiments often requires jumping between browser tabs and CLI tools. This skill enables direct operation of Dingo SaaS through natural language by mapping requests to backend API actions.

What it does

• Dataset Management Handles listing, creating, and uploading local files to Dingo datasets.
• Experiment Control Starts, stops, and schedules evaluation experiments with specific task types and evaluators.
• Report Analysis Retrieves execution logs, summaries, and result rows for completed runs.
• Metric Configuration Manages metric groups and custom LLM-based metrics, including drafting and testing new criteria.
• Automated Workflows Chains multiple API calls, such as uploading a file then immediately initializing a dataset.

How it works

1. Authentication Set your DINGO_SAAS_API_KEY environment variable or use the built-in configure command to save it to a local config file.
2. Intent Mapping State your goal, such as "run experiment Alpha with the latest dataset."
3. Execution The skill identifies the correct module scripts and executes Node.js commands to interact with the REST API.
4. Resolution If an ID is missing, it automatically searches by name to find the correct resource before acting.

Frameworks & tools

This skill uses Node.js scripts to interface with the Dingo SaaS REST API. It manages local configuration files and handles multipart file uploads for dataset ingestion.

Why this beats prompting it yourself

Writing prompts to interact with a complex API often results in invalid JSON payloads or missed dependency steps. This skill has hardcoded knowledge of the Dingo SaaS schema, ensuring that multi-step operations like file-to-dataset conversion or experiment scheduling follow the exact backend requirements every time.

Use cases

• Upload a local JSONL file and create a new dataset in one command.
• Schedule an existing experiment to run every Monday and Wednesday at 9:00 AM.
• Monitor a running report's logs in real-time and summarize failures.
• Draft and "quick try" a custom LLM metric before saving it to a group.

Known limitations

Requires Node.js installed on the host machine. It cannot interact with unimplemented Dingo SaaS pages or administrative global settings.