Full Arboricx parsing in tricu

notes/tricu-cli-debugging.md

@@ -0,0 +1,17 @@
+# tricu CLI debugging notes
+
+For ad-hoc expressions, prefer stdin mode and set `TRICU_DB_PATH` to a DB that already has library definitions imported:
+
+```sh
+TRICU_DB_PATH=/tmp/gpt.db ./result/bin/tricu eval -t decode <<'EOF'
+main = <expression-to-run>
+EOF
+```
+
+Important details:
+
+- `eval` from stdin evaluates the submitted program and uses its final/main result.
+- When using `-f FILE`, the CLI expects a `main` definition in the evaluated file context.
+- With `TRICU_DB_PATH=/tmp/gpt.db`, definitions already loaded into that content store are in scope; do not add `!import` lines unless you intentionally want file import preprocessing.
+- `!import "lib/arboricx.tri" !Local` is relative to the file being preprocessed; from temp files it will look under `/tmp`, so avoid that pattern for scratch files.
+- Do not inspect huge Arboricx values with `-t fsl`; write small predicates/accessors and return booleans, numbers, or byte strings decoded with `-t decode`.