[BornPsych]

Reads the single program.mavros that Mavros now writes (reilabs/mavros#429), instead of basic.json + r1cs.bin.

1. prepare --compiler mavros <program.mavros>: one input, the --r1cs flag is gone. A package directory also works and resolves to its target/program.mavros, where mavros compile writes by default.
2. mavros-artifacts checks the file on load (digests, manifest, bytecode header); the ABI and bytecode come out of the same file. A file that fails a check is refused with the fix: recompile with the pinned Mavros, or with --field bn254 for an artifact over another field.
3. Proofs bind the Mavros circuit digest (R1CS + manifest) into the transcript, so a codegen-only recompile keeps existing verifiers valid. verifier_is_bound_to_the_mavros_digest covers it.
4. prove returns an error instead of panicking in the VM when a key holds a program of another format version.
5. The power fixture is regenerated with that CLI; README and scripts/mavros_parity.sh follow.

Mavros is pinned to the head of reilabs/mavros#429 (c445d5d); repin to its merge commit once that lands.

The Mavros witness layout gains logup_repetitions, which changes the Mavros prover and scheme records. That is recorded under key format 4.0 rather than a new version, since 4.0 has not shipped: Mavros keys prepared on main since #507 must be regenerated.

Regenerate fixtures with mavros compile <dir> --output <dir>/program.mavros.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
docs-provekit-org	Ready	Preview	Nov 11, 2026 8:09pm UTC

[github-actions]

CSP benchmarks

Metric	Value
Workflow status	[PASS] success
Commit	278f505b0aed
Run	#43093856168
Circuits benchmarked	21
Iterations averaged per circuit	3

Prover time, peak RSS, peak heap, and verifier time are arithmetic means across the iterations. Peak heap comes from the largest peak memory entry in provekit-cli prove's tracing output; peak RSS is reported by /usr/bin/time -v (max-resident-set-size).

Each metric cell shows the current value followed by the percentage delta against the latest successful main run #43001011490. (new) marks circuits absent from the baseline.

Results

Circuit	Constraints	Witnesses	Prover time	Peak RSS	Peak heap	Verifier time	Proof size	PKP size
ecdsa_p256	143,321 (±0.0%)	258,215 (±0.0%)	1.40 s (-1.9%)	267 MB (+0.7%)	232 MB (±0.0%)	107 ms (-3.0%)	447 KB (-1.5%)	831 KB (±0.0%)
keccak_1024	788,814 (±0.0%)	1,490,812 (±0.0%)	3.96 s (-0.8%)	1.05 GB (+0.2%)	1.05 GB (±0.0%)	477 ms (-1.4%)	484 KB (-1.5%)	6.01 MB (±0.0%)
keccak_128	157,522 (±0.0%)	304,529 (±0.0%)	963 ms (-2.0%)	281 MB (+0.7%)	248 MB (±0.0%)	120 ms (±0.0%)	449 KB (-0.9%)	1.24 MB (±0.0%)
keccak_2048	1,507,374 (±0.0%)	2,840,532 (±0.0%)	7.72 s (-1.0%)	1.99 GB (+0.1%)	2.07 GB (±0.0%)	913 ms (-2.5%)	544 KB (-0.2%)	12.15 MB (±0.0%)
keccak_256	247,782 (±0.0%)	474,010 (±0.0%)	1.16 s (-2.2%)	334 MB (+0.6%)	304 MB (±0.0%)	170 ms (±0.0%)	451 KB (+1.2%)	1.89 MB (±0.0%)
keccak_512	428,126 (±0.0%)	812,944 (±0.0%)	2.08 s (-0.8%)	586 MB (+0.3%)	558 MB (±0.0%)	270 ms (-2.4%)	474 KB (+0.5%)	3.25 MB (±0.0%)
poseidon2_12	479 (±0.0%)	563 (±0.0%)	40 ms (-20.0%)	15.56 MB (+16.8%)	4.17 MB (±0.0%)	10 ms (±0.0%)	227 KB (-0.2%)	437 KB (±0.0%)
poseidon2_16	556 (±0.0%)	719 (±0.0%)	47 ms (±0.0%)	16.21 MB (+16.1%)	4.33 MB (+0.1%)	10 ms (±0.0%)	227 KB (+0.7%)	530 KB (±0.0%)
poseidon2_2	231 (±0.0%)	278 (±0.0%)	40 ms (±0.0%)	14.82 MB (+18.6%)	3.44 MB (±0.0%)	10 ms (±0.0%)	227 KB (+1.4%)	108 KB (±0.0%)
poseidon2_4	529 (±0.0%)	535 (±0.0%)	30 ms (-18.3%)	15.17 MB (+17.5%)	3.63 MB (±0.0%)	10 ms (±0.0%)	227 KB (+2.0%)	31.93 KB (±0.0%)
poseidon2_8	363 (±0.0%)	423 (±0.0%)	40 ms (±0.0%)	15.32 MB (+15.1%)	4.08 MB (±0.0%)	10 ms (±0.0%)	229 KB (+1.6%)	365 KB (±0.0%)
poseidon_12	504 (±0.0%)	524 (±0.0%)	40 ms (-7.6%)	16.02 MB (+17.6%)	4.27 MB (±0.0%)	10 ms (±0.0%)	224 KB (-0.3%)	569 KB (±0.0%)
poseidon_16	609 (±0.0%)	633 (±0.0%)	50 ms (±0.0%)	16.77 MB (+15.8%)	4.93 MB (+0.1%)	20 ms (±0.0%)	231 KB (+1.6%)	717 KB (±0.0%)
poseidon_2	240 (±0.0%)	249 (±0.0%)	30 ms (-18.3%)	14.75 MB (+19.2%)	3.40 MB (±0.0%)	10 ms (±0.0%)	225 KB (-0.5%)	59.38 KB (±0.0%)
poseidon_4	297 (±0.0%)	309 (±0.0%)	40 ms (±0.0%)	15.74 MB (+16.9%)	4.02 MB (±0.0%)	10 ms (±0.0%)	220 KB (-2.2%)	333 KB (±0.0%)
poseidon_8	402 (±0.0%)	418 (±0.0%)	40 ms (±0.0%)	15.73 MB (+15.8%)	4.15 MB (±0.0%)	10 ms (±0.0%)	221 KB (-2.5%)	448 KB (±0.0%)
sha256_1024	196,913 (±0.0%)	339,721 (±0.0%)	1.07 s (-1.2%)	321 MB (+0.6%)	286 MB (±0.0%)	180 ms (-1.8%)	453 KB (+0.1%)	1.87 MB (-0.4%)
sha256_128	46,371 (±0.0%)	80,931 (±0.0%)	270 ms (-2.4%)	86.55 MB (+1.5%)	67.81 MB (±0.0%)	50 ms (±0.0%)	386 KB (-1.4%)	498 KB (-0.2%)
sha256_2048	345,372 (±0.0%)	612,681 (±0.0%)	2.07 s (-1.3%)	569 MB (+0.3%)	535 MB (±0.0%)	323 ms (±0.0%)	463 KB (-2.6%)	3.07 MB (+0.8%)
sha256_256	67,877 (±0.0%)	117,901 (±0.0%)	443 ms (-2.9%)	142 MB (+1.5%)	120 MB (±0.0%)	70 ms (±0.0%)	406 KB (+0.1%)	698 KB (+0.4%)
sha256_512	110,889 (±0.0%)	191,841 (±0.0%)	570 ms (-1.7%)	177 MB (+2.3%)	153 MB (±0.0%)	100 ms (-9.1%)	403 KB (-0.9%)	1.09 MB (-0.3%)

[zkfriendly]

LGTM!

One note is that logup_repetitions is newly added to the WitnessLayout without bumping the PROVER_VERSION. Since this is a stack PR it's fine if we merge the stack all together.