Epistemic Protocols is a plugin for Claude Code that inserts decision checkpoints—before irreversible actions, when multiple approaches exist, or when intent might be misread.
Three protocols:
- /lens(/prothesis) — before analysis, lets you choose the perspective (performance, ops complexity, data model fit, etc.)
- /gap(/syneidesis) — at decision points, surfaces potential gaps as questions
- /clarify(/hermeneia) — when your request might be ambiguous, helps refine intent
Unknown unknowns:
Even well-scoped requests hide blind spots—the unknown unknowns. Epistemic Protocols help surface them early:
/lens broadens the frame so you don’t miss critical perspectives; /gap turns “things we didn’t think to ask” into concrete questions; /clarify reduces ambiguity that masks hidden assumptions.
In short, they convert unknown unknowns into known considerations when change is still cheap.
Principle: the AI illuminates options; you choose. Recognition over recall—selecting from clear choices beats guessing.
GitHub: https://github.com/jongwony/epistemic-protocols
Example:
``` > /lens "Should I use Redis or PostgreSQL for this cache?" ```
Claude shows lenses (Performance, Ops complexity, Data model fit…). You pick what matters, then it analyzes through those lenses.
Show HN: Epistemic Protocols – Decision Checkpoints for Claude Code
github.com