01Interoperable Interfaces: Provides a command-line interface (CLI) and MCP (Managed Code Protocol) server with provider/tool adapters to avoid vendor lock-in across AI tools.
02LLM-Agent-First Design: Defaults to structured JSON output and uses precise exit codes, with human-readable formatting available via a `--human` flag.
03Anti-Hallucination Protection: Employs four layers of validation (Schema, Semantic, Cross-File, State Machine) to prevent AI-generated errors and ensure data integrity.
04Portable Memory & Provenance: Organizes tasks in a Project -> Epic -> Task hierarchy, maintaining immutable audit trails and tracing every artifact to its origin.
05Stable Task IDs: Uses flat, sequential, and eternal IDs (e.g., T001) that decouple identity from hierarchy, ensuring external references remain valid forever.
06104 GitHub stars
