[TimothyBanks]

High Level Overview of Change

Hooks up binaryen wasm-opt during the build process to optimize wasm binary sizes.

Contract	Before	After	Saved
keylet_exists	16,520	10,645	35%
host_functions_test	14,751	10,530	28%
trace_escrow_ledger_object	14,193	9,646	32%
trace_escrow_finish	14,011	10,064	28%
trace_escrow_account	13,852	9,591	30%
atomic_swap1	8,445	5,985	29%
atomic_swap2	7,011	5,054	27%
float_tests	6,454	4,278	33%
gas_benchmark	5,793	4,213	27%
nft_owner	4,827	3,176	34%
freelancer_escrow	4,396	2,851	35%
oracle	4,277	2,866	32%
kyc	1,597	943	40%
ledger_sqn	1,006	376	62%
notary	927	561	39%
hello_world	456	217	52%

From e2e tests:

🗜️  Optimizing WASM with wasm-opt version 116 (version_116)...
   ✅ atomic_swap1.wasm: 8445 -> 5985 bytes (29% smaller)
   ✅ atomic_swap2.wasm: 7011 -> 5054 bytes (27% smaller)
   ✅ freelancer_escrow.wasm: 4396 -> 2845 bytes (35% smaller)
   ✅ hello_world.wasm: 456 -> 207 bytes (54% smaller)
   ✅ kyc.wasm: 1597 -> 933 bytes (41% smaller)
   ✅ ledger_sqn.wasm: 1006 -> 366 bytes (63% smaller)
   ✅ nft_owner.wasm: 4827 -> 3170 bytes (34% smaller)
   ✅ notary.wasm: 927 -> 551 bytes (40% smaller)
   ✅ oracle.wasm: 4277 -> 2868 bytes (32% smaller)
   ✅ float_tests.wasm: 6454 -> 4272 bytes (33% smaller)
   ✅ gas_benchmark.wasm: 5793 -> 4208 bytes (27% smaller)
   ✅ host_functions_test.wasm: 14751 -> 10546 bytes (28% smaller)
   ✅ keylet_exists.wasm: 16520 -> 10641 bytes (35% smaller)
   ✅ trace_escrow_account.wasm: 13852 -> 10311 bytes (25% smaller)
   ✅ trace_escrow_finish.wasm: 14011 -> 10511 bytes (24% smaller)
   ✅ trace_escrow_ledger_object.wasm: 14193 -> 9687 bytes (31% smaller)
🗜️  wasm-opt: 16 contracts, 118516 -> 82155 bytes total

Context of Change

Type of Change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Refactor (non-breaking change that only restructures code)
• Tests (added tests for existing code, or tests for the new feature in this PR)
• Documentation update
• Build / CI / tooling change
• Release

[ripple-code-reviewer]

The main functional addition is scripts/wasm-opt.sh plus wiring it into build.sh/setup.sh/CI. The wasm-opt.sh tool logic (resolution, install, hashing, stamp skip) looks sound, but the way it's invoked unconditionally at the end of build.sh conflicts with its own hard-fail-on-missing-directory behavior for non-release builds.

[ripple-code-reviewer]

Solid, well-documented addition of a wasm-opt post-processing step. The script itself is careful (pinned tool version, deterministic MVP-feature flags, explicit vs default target handling). Two things worth a second look: the opt-stamp cache only keys off the wasm file's own hash, so changing WASM_OPT_FLAGS or bumping the pinned wasm-opt version won't force re-optimization of files that cargo didn't rebuild — stale bytes could silently ship or skew benchmark-gas.sh numbers. Also, the new 'Ensure wasm-opt' step added to the UI/docker job in test.yml sits right after an npm-only setup with no visible Rust/cargo caching in this hunk — if that job has no cargo toolchain and no cached wasm-opt binary, the auto-install path will fail the job.

[sappenin]

@TimothyBanks Can you take a look at the two bot comments (here and here) and maybe have claude rectify those?

[sappenin]

See my main comment about addressing the PR bot's suggestions. Otherwise LGTM

[TimothyBanks]

@sappenin Should be addressed with my last commit.

[ripple-code-reviewer]

This adds a wasm-opt.sh script wired into build.sh/setup.sh/CI to shrink release WASM artifacts with Binaryen. The script is well-structured (tool resolution → install fallback → optimize with a stamp-based skip cache), and the CI/action-pinning changes are standard SHA-pinning. The one real issue is that the stamp file only tracks the wasm binary's content hash, not the flags or wasm-opt version used to produce it, so changing WASM_OPT_FLAGS (or the pinned crate version) between runs without an intervening cargo build will silently skip re-optimization.

[ripple-code-reviewer]

This MR wires binaryen's wasm-opt into the build pipeline (build.sh, setup.sh, CI workflows) via a new scripts/wasm-opt.sh, plus pins GitHub Actions to SHAs. The new script itself (idempotency via hash+flags stamp, mandatory-tool vs no-op-when-nothing-to-optimize, MVP-feature flag rationale) is well thought out and matches the accompanying documentation. The main gap is that tool resolution never verifies the found binary's version against the pinned WASM_OPT_CRATE_VERSION, which combined with CI's cache key (based only on Cargo.lock, not on the wasm-opt pin) means a future pin bump can silently keep using a stale cached binary in CI.

[ripple-code-reviewer]

This is a well-engineered change: the new scripts/wasm-opt.sh is carefully written with sensible fallback/pin-verification logic, and the CI/doc updates (action SHA-pinning, wasm-opt wiring into build.sh/docs.sh/setup.sh, and the AGENTS.md rationale) are internally consistent with the script's actual behavior. I didn't find correctness, security, or resource-leak bugs in the added code. The only note is a minor, purely cosmetic log-message issue in build.sh.

[sappenin]

@TimothyBanks I don't think the e2e tests are failing in main (but I can't seem to find a recent build to confirm that). The e2e-tests are passing locally for me in main, FWIW.

Can you look at that one before merging?

[ripple-code-reviewer]

This change wires up Binaryen's wasm-opt into the build pipeline (build.sh, setup.sh, CI workflows) via a new scripts/wasm-opt.sh, plus pins GitHub Actions to commit SHAs and documents the new behavior extensively in AGENTS.md/README. The new script is carefully engineered — it guards against div-by-zero in the size-delta calc, correctly exempts user-set/native wasm-opt from the version-pin check via a -ef identity test, handles the no-release-WASM-found case distinctly from an explicitly-named-but-missing target, and the mandatory-tool / in-place-write tradeoffs are well justified in AGENTS.md. I did not find any high-confidence correctness or security bugs introduced by this diff; the CI action SHA pins can't be verified against upstream tags without network access, so I'm not flagging them per the project's own guidance to avoid unverifiable configuration claims.

[ripple-code-reviewer]

This PR adds scripts/wasm-opt.sh (a Binaryen wasm-opt wrapper that shrinks release WASM artifacts in place), wires it into build.sh/setup.sh/CI, and pins third-party GitHub Actions to full commit SHAs. The new shell script is unusually well-engineered: it handles tool resolution order (explicit WASM_OPT > PATH > cargo bin), version-pins the cargo-installed binary while exempting user-supplied/system binaries, guards against div-by-zero in the size-delta calculation, cleans up temp files on failure, and distinguishes 'no release WASM to optimize' (exit 0) from 'explicitly named target missing' (exit 1). I traced the control flow (resolve_wasm_opt, the pipefail/head interaction in binaryen_version_of, the -e semantics around the if-blocks in install_pinned_wasm_opt) and didn't find a case that produces wrong behavior. The AGENTS.md documentation additions are detailed and internally consistent with the script and CI wiring (e.g., --ensure-tool calls in setup.sh/docs.yml/test.yml, SKIP_WASM_OPT/WASM_OPT_NO_INSTALL env vars, cache-staleness handling for version bumps). The GitHub Actions SHA-pinning changes look like good practice; I can't independently verify each 40-char SHA corresponds to the stated tag, so I'm not flagging those without evidence of a mismatch. No correctness, security, or resource-leak issues found in the added code that meet the bar for confident flagging.

[TimothyBanks]

@sappenin Fixed