- LOC: ~139,200 (monorepo with many sub-standards)
- Languages: ReScript, Idris2, Agda, Rust, Haskell, Zig, Nickel
- Existing ABI proofs: Multiple
src/abi/*.idracross sub-projects (a2ml, axel-protocol, avow-protocol, lol, etc.) - Existing verification:
lol/proofs/theories/information_theory.agdawith A2ML Idris2 proofs ina2ml/src/A2ML/Proofs.idr - Dangerous patterns:
lol/proofs/theories/information_theory.agda: 6postulate(information theory axioms)axel-protocol/src/Tea.resandAxelApp.res:Obj.magicfor DOM operationslol/src/abi/Locale.idr: mentions avoiding believe_me (good practice)
information_theory.agdahas 6 postulated axioms about entropy, mutual information, etc.- Audit: which are genuine mathematical axioms vs. provable lemmas?
- Entropy non-negativity and chain rule should be constructively provable
Proofs.idrexists — audit completenessParser.idr,TypedCore.idr,Surface.idr— prove parser/type-system correspondence- A2ML is a markup format standard — parser correctness ensures documents are faithfully processed
avow-protocol/avow-lib/src/abi/Consent.idr,Unsubscribe.idr- Consent management is GDPR-relevant — prove consent state transitions are correct
- Prove: unsubscribe always terminates in a non-consented state
axel-protocol/src/Tea.res— 8+Obj.magiccalls for DOM rendering- Lower priority than the protocol specification proofs
groove-protocol/reference/groove-proxy/GrooveProxy.idr— Idris2 reference implementation- Prove: proxy faithfully implements the Groove protocol specification
- Agda for information theory (extend existing proofs, eliminate postulates)
- Idris2 for A2ML, Avow, Groove protocol correctness (already in use)
MEDIUM — Standards monorepo. The LOL information theory postulates and Avow consent proofs are the highest-value targets. A2ML proofs already exist and need completion audit.
Template ABI removed -- was creating false impression of formal verification. The removed files (Types.idr, Layout.idr, Foreign.idr) contained only RSR template scaffolding with unresolved {{PROJECT}}/{{AUTHOR}} placeholders and no domain-specific proofs.