Palantir explicitly structures AI FDE around two tiers: Modes (the broad task at hand, e.g., data integration or ontology editing) and Skills (granular capabilities usable across multiple modes). This GitHub repository library mirrors this exact two-tier hierarchy — skill.md maps to Palantir’s Skills layer, agents.md maps to full Mode-level agent configurations. Your prompt library will sit as the execution layer beneath both.
https://github.com/s-andthat/palantir-ai-fde-library
Hopefully someone will find this useful.