chore(loc): exclude fuzz and guest programs from LOC count

[diegokingston]

The loc tool counted executor/programs (RISC-V guest test programs) as production code. Add fuzz, programs and program_artifacts to the EXCLUDED patterns.

This removes 502 LOC of guest test programs (executor/programs/rust/*, 30 files) from the reported total (32,937 -> 32,435). Fuzz harnesses already scored 0 because fuzz_targets/ coincidentally matched target; they are now excluded explicitly so a harness placed elsewhere under crypto/math/fuzz/ is not counted either.

[github-actions]

Codex Code Review

No issues found in the PR diff.

The changes only add the MIT LICENSE, a short Fiat-Shamir README, and LOC-tool exclusion patterns for fuzz/program fixture directories. I did not see security, correctness, or significant performance concerns in these changes.

Tests not run; this was a diff-only review.

[claude]

Review: chore(loc): exclude fuzz and guest programs from LOC count

Overall: LGTM with one observation.

Observation — dual semantics of EXCLUDED

The EXCLUDED slice is used in two distinct ways inside main.rs:

1. Exact string match at line 42: EXCLUDED.contains(&crate_name) — compares the literal top-level directory name against slice entries. Glob wildcards have no effect here.
2. Glob patterns at line 61: passed to tokei's get_statistics — wildcards do work here.

The new patterns (*fuzz*, *programs*, *program_artifacts*) only take effect through tokei, which is the desired behaviour (they suppress subdirectories found recursively during counting). They never fire in the contains() check, but that check only guards top-level crate dirs and none is literally named programs or fuzz, so there is no functional problem.

The existing code already relied on this split — not introduced here. Worth a follow-up to rename the constant (e.g. EXCLUDED_GLOBS) to make the two call sites obviously different, but not blocking.

Bundled changes

LICENSE and the Fiat-Shamir README are unrelated to the LOC fix. No issues with the content.

[github-actions]

Codex Code Review

No issues found in the PR diff.

The change only adds LOC-report exclusions for fuzz harnesses, guest programs, and program artifacts. I don’t see security, correctness, or significant performance concerns from these added patterns. Tests were not run; review was based on the diff and matching repo paths.

[claude]

Review: chore(loc): exclude fuzz and guest programs from LOC count

The change is correct and consistent with the existing exclusion convention in this file. One thing worth noting:

Low — *programs* is a broad glob

The pattern *programs* will silently exclude any future crate whose name contains that word (e.g. syscall_programs, programs_api). The existing patterns (*bench*, *tests*, etc.) follow the same convention, so this is consistent — but those words are less likely to appear in production crate names than "programs" might be in a zkVM project.

If there's ever a production crate whose path includes "programs", it'll be dropped from the LOC count with no warning. A narrower path like *guest_programs* or executor/programs would scope it more tightly if that risk matters.

Low — *program_artifacts* has no current target

No directory matching this pattern exists in the repo today. Harmless as a forward-looking exclusion, but worth noting it's speculative.

Otherwise no issues — no security, performance, or correctness concerns.