tooling improvements

docs/cli.md

@@ -65,6 +65,7 @@ pit build -b debug                # build type: release (default), debug, minsiz
 pit build --list-targets          # list available targets
 pit build --force                 # force rebuild
 pit build --dry-run               # show what would be built
+pit build --verbose               # print resolved flags, commands, cache status
 ```
 
 ### pit test
@@ -330,9 +331,14 @@ pit mcode <file.cm>
 Apply the full optimization pipeline to a source file and output optimized mcode as JSON.
 
 ```bash
-pit streamline <file.cm>
+pit streamline <file.cm>                # full optimized IR as JSON (default)
+pit streamline --stats <file.cm>        # summary stats per function
+pit streamline --ir <file.cm>           # human-readable IR
+pit streamline --check <file.cm>        # warnings only (e.g. high slot count)
 ```
 
+Flags can be combined. `--stats` output includes function name, args, slots, instruction counts by category, and nops eliminated. `--check` warns when `nr_slots > 200` (approaching the 255 limit).
+
 ### pit qbe
 
 Compile a source file to QBE intermediate language (for native code generation).
@@ -367,6 +373,17 @@ Ahead-of-time compile and execute a program natively.
 pit run_aot <program.ce>
 ```
 
+### pit seed
+
+Regenerate the boot seed files in `boot/`. Seeds are pre-compiled mcode IR (JSON) that bootstrap the compilation pipeline on cold start. They only need regenerating when the pipeline source changes in a way the existing seeds can't compile, or before distribution.
+
+```bash
+pit seed                # regenerate all boot seeds
+pit seed --clean        # also clear the build cache after
+```
+
+The engine recompiles pipeline modules automatically when source changes (via content-addressed cache). Seeds are a fallback for cold start when no cache exists.
+
 ### Analysis
 
 ### pit explain