Early benchmarks: 30%+ input-token reduction from reusable reasoning — read the field note
Explore the spec Explore the spec

Playground

Live editor + validator.

Edit the Scholialang .srml source on the left. The AST on the right re-renders after every pause, showing each atom's v0.6 content-addressable canonical_id. The validator below tells you which v0.6 validity rules the current trace satisfies, including non-fatal warnings and the new canonical_id_well_formed check.

Scholialang source (.srml)

Parsed AST

Validator — SCHOLIA_v0.6_SPEC §4 rules 1–18 (incl. canonical_id_well_formed) + warnings
    Lazy prelude preview — the 3 core modes (hash_only · hash_list · inline)

    How a later session would reuse this trace's atoms by canonical_id instead of replaying their full XML. Pick a render mode; relative size shows the token tradeoff. The two experimental quality-recovery arms (§14) are excluded — they are not part of the finalized v0.6 contract.

    This is an illustrative parser. The canonical implementation is the Python rsi-scholia-core module defined in EPIC-018 — still in flight. Parse errors here should not be treated as authoritative.