JSON Harbor is an interactive browser-based learning game that teaches JSON syntax, schema validation, structural rules, and data transformation through progressive dock missions.

reads hostnames from spreadsheet and changes JSON attributes and creates PDF files as final

No description available

I created a Docker-based AI evaluation task using Harbor framework. The task required processing CSV data into filtered and sorted JSON output. I defined task metadata, container environment, reference solution, and deterministic test validation. The task was validated using Oracle agent (expected solution) and NOP agent (baseline).

A standardized data processing task built for the Harbor evaluation framework. This task requires an AI agent to filter active user records from a JSON dataset within a containerized Python 3.13 environment. Includes automated Oracle (1.0) and NOP (0.0) validation suites.

Harbor evaluation task that cleans whitespace from JSON fields using Docker and jq.

No description available

No description available

A Harbor benchmark task that evaluates an agent's ability to read a JSON file, compute basic statistics, and write formatted output.