pattern13_prml_commit_reveal.py makes the co-authored Pattern 13 executable end to end:
PRML pre-registration + a valichord_attestation-style bundle + a local commit-reveal round,
then three adversarial cases showing each layer catches what the others can't.
python3 pattern13_prml_commit_reveal.py
Honesty: the PRML layer uses the real falsify_prml reference; the bundle hashing
(Merkle root, bundle_hash/content_hash) is real crypto matching the v1.2 format; the
commit-reveal is real crypto but simulated locally (production ValiChord runs on
Holochain across isolated nodes). No network, stdlib only.
lm_eval_to_prml.py turns an lm-evaluation-harness run into a pre-registered PRML claim,
the honest way: lock the bar from the task config BEFORE the run, verify the observed
metric from results.json AFTER. Avoids the "hash after the run" anti-pattern (anti/A1).
python3 lm_eval_to_prml.py # built-in demo
python3 lm_eval_to_prml.py --mode lock --results results.json \
--task hellaswag --metric acc_norm --threshold 0.75 # real run
Real: PRML canonicalisation/hashing/verdicts (falsify_prml — the same reference the spec and the
4 byte-equivalent impls use, no parallel schema). Sample-modelled: a faithful in-file lm-eval results
dict so it runs with no lm-eval install; --results accepts a real one.
This is the flagship reference integration — what it looks like to drop a pre-committed eval claim
into a tool a platform already uses (lm-evaluation-harness, the most-used LLM harness). The demo's
[ATTEST] step emits the locked claim as an in-toto / ITE-6 Statement (falsify >= 0.3.8), so a
host that already ingests SLSA/in-toto can treat a pre-registered eval as one more predicate type — the
3-function embed path is in docs/EMBED.md.
Related upstream thread: lm-evaluation-harness PR #3752.