This one needs the most work:

1. run.sh has hardcoded path: /home/lind/lind-wasm/src/wasmtime/target/debug/wasmtime — won't work outside that specific container. Should use lind_run or be parameterized.

2. clean.sh references $SYSROOT which is never defined in that script — will fail with "unbound variable" under set -eu.

3. clean.sh uses sudo rm -rf — dangerous in a build script. The build shouldn't require sudo.

4. Modifies shared sysroot — build_python_wasm.sh creates empty placeholder .a files (libwasi-emulated-getpid.a, etc.) directly in $SYSROOT/lib/wasm32-wasi/. This modifies the shared sysroot and affects all other builds. Should use the overlay pattern instead.

5. Hardcodes CLANG_BIN instead of loading from .toolchain.env like other scripts.

6. make ... &> make.log — hides all build output. If the build fails, you have to dig into the log file to find the error. Other scripts use V=1 to show output.

7. run_tests.sh modifies the build Makefile with sed — very fragile. Will break if the Makefile format changes.

8. Missing space before || in lind-boot precompile call.

9. build_shared_python_wasm.sh duplicates the same 50-line dylink pipeline from PR #185 — should use a shared helper.

10. Duplicate --export flags in the shared python link command — --export=__wasm_call_ctors, --export-if-defined=__wasm_init_tls, and --export=__tls_base each appear twice.

Cross-cutting feedback for all the dynamic PRs (#185, #188, #189, #190, #191):

The dylink pipeline (clang link → add-export-tool ×3 → wasm-opt → precompile → stage) should be extracted into a shared helper script. It's duplicated 8+ times across these PRs with only the library name changing. A single scripts/build_dynamic_lib.sh taking the static archive and output name as args would eliminate ~400 lines of duplication and make it much easier to update the pipeline when flags change.