diff --git a/PLAN.md b/PLAN.md
index 4847b25..dca3ead 100644
--- a/PLAN.md
+++ b/PLAN.md
@@ -173,21 +173,120 @@ Bux is a fast, compiled, strongly-typed, multi-paradigm systems programming lang
 
 ---
 
-## Phase 7 — Self-Hosting: The Great Rewrite (Week 19-26)
+## Phase 6.5 — Self-Hosting Audit (Completed 2026-05-31)
+
+### Source File Analysis
+
+| File | Lines | Procs | Complexity | Bux Readiness |
+|------|-------|-------|------------|---------------|
+| `source_location.nim` | 8 | 0 | Trivial struct | ✅ Ready |
+| `main.nim` | 6 | 0 | CLI entry | ✅ Ready |
+| `scope.nim` | 47 | 4 | Simple | ✅ Ready |
+| `manifest.nim` | 79 | 2 | TOML parser | ⚠️ Needs TOML/INI parser |
+| `hir.nim` | 184 | 0 | Type defs | ✅ Ready |
+| `types.nim` | 185 | 44 | Factories | ✅ Ready |
+| `token.nim` | 305 | 12 | Enum + helpers | ✅ Ready |
+| `cli.nim` | 390 | 15 | File I/O, process | ⚠️ Needs File I/O, path ops |
+| `ast.nim` | 400 | 6 | Complex case-object | ✅ Ready (algebraic enums) |
+| `c_backend.nim` | 519 | 16 | Code generation | ⚠️ Needs String formatting |
+| `lexer.nim` | 567 | 37 | State machine | ⚠️ Needs String split/compare |
+| `sema.nim` | 892 | 27 | Type checking | ⚠️ Needs Table[String,...] |
+| `parser.nim` | 1220 | 81 | Pratt parser | ⚠️ Needs seq/array ops |
+| `hir_lower.nim` | 1233 | 29 | Tree transform | ⚠️ Needs Table, HashSet |
+
+### Nim Patterns → Bux Equivalents
+
+| Nim Pattern | Used In | Bux Status |
+|-------------|---------|------------|
+| `Table[string, T]` | sema, hir_lower, c_backend (23 uses) | ❌ **Blocker** — need `StringMap<V>` |
+| `HashSet[string]` | hir_lower (1 use) | ❌ Can use `StringMap<bool>` workaround |
+| `seq[T]` with push/len/iter | All files (200+ uses) | ⚠️ `Array<T>` exists, needs richer API |
+| `&"..."` / `fmt"..."` | sema, c_backend (119 uses) | ❌ **Blocker** — need string formatting |
+| `split()`, `join()` | lexer, parser, cli | ❌ **Blocker** — need String split/join |
+| `case obj.kind of...` | All files (90+ uses) | ✅ `match` with algebraic enums |
+| `for x in collection` | All files (200+ uses) | ✅ Supported |
+| `var` parameters | Multiple | ✅ Use pointers (`*T`) |
+| File read/write | cli | ❌ Need `readFile`, `writeFile` |
+| OS path operations | cli, manifest | ❌ Need path join, exists |
+
+### Rewrite Order (Dependency-driven)
+
+```
+Phase 7.0 — Stdlib blockers:
+  ├── StringMap<V> (blocker #1 — needed by all modules)
+  ├── String split/join (blocker #2 — needed by lexer, parser)
+  ├── String formatting (blocker #3 — needed by sema, c_backend)
+  ├── File I/O (readFile, writeFile, fileExists)
+  └── OS path (joinPath, parentDir)
+
+Phase 7.1 — Foundation (no internal deps):
+  ├── token.bux (enum + helpers)
+  ├── source_location.bux (struct)
+  ├── types.bux (enum + factories)
+  ├── scope.bux (symbol table — needs StringMap)
+  └── hir.bux (type definitions)
+
+Phase 7.2 — Frontend (depends on 7.1):
+  ├── lexer.bux (needs String split/compare)
+  ├── ast.bux (algebraic enums)
+  └── parser.bux (Pratt parser, needs Array<T>)
+
+Phase 7.3 — Analysis (depends on 7.2):
+  ├── sema.bux (type checking, needs StringMap, formatting)
+  └── manifest.bux (TOML parser)
+
+Phase 7.4 — Backend (depends on 7.3):
+  ├── hir_lower.bux (tree transform, needs StringMap, HashSet)
+  └── c_backend.bux (code gen, needs String formatting)
+
+Phase 7.5 — Driver (depends on all):
+  ├── cli.bux (file I/O, argument parsing)
+  └── main.bux (entry point)
+```
+
+### Risk Assessment
+
+| Risk | Severity | Mitigation |
+|------|----------|------------|
+| StringMap not working for String keys | **High** | Already have working `StringMap<V>` in stdlib using strcmp |
+| `&key as *void` precedence bug | **Medium** | Workaround: use intermediate `*K` variable |
+| Cross-module generics not working | **Medium** | All compiler code will be in one package (merged via stdlib mechanism) |
+| `Map_Len` monomorphization bug | **Low** | Avoid calling `Map_Len` with explicit type args; inline the body |
+| String formatting complexity | **Medium** | Use StringBuilder pattern instead of printf-style formatting |
+| Array<T> API gaps | **Low** | Extend Array module as needed during porting |
+
+### Estimated Effort
+
+| Phase | Bux LOC | Effort |
+|-------|---------|--------|
+| 7.0 Stdlib blockers | ~300 | 1-2 sessions |
+| 7.1 Foundation | ~600 | 1-2 sessions |
+| 7.2 Frontend | ~1800 | 3-4 sessions |
+| 7.3 Analysis | ~900 | 2-3 sessions |
+| 7.4 Backend | ~1700 | 3-4 sessions |
+| 7.5 Driver | ~400 | 1 session |
+| **Total** | **~5700** | **11-16 sessions** |
+
+---
+
+## Phase 7 — Self-Hosting: The Great Rewrite
 
 **Goal:** Bux compiler compiles itself. This is the **main milestone**.
 
-| Task | Details |
-|------|---------|
-| `7.1` Port lexer | Rewrite `lexer.nim` → `Lexer.bux` |
-| `7.2` Port parser | Rewrite `parser.nim` → `Parser.bux` |
-| `7.3` Port sema | Rewrite `sema.nim` → `Sema.bux` |
-| `7.4` Port HIR | Rewrite `hir.nim` → `Hir.bux` |
-| `7.5` Port LIR | Rewrite `lir.nim` → `Lir.bux` |
-| `7.6` Port C backend | Rewrite `c_backend.nim` → `CBackend.bux` |
-| `7.7` Port CLI | Rewrite `main.nim` → `Main.bux` |
-| `7.8` Dogfooding | Use `buxc` (Nim) to build `buxc2` (Bux). Then use `buxc2` to build `buxc3`. Compare bit-for-bit. |
-| `7.9` Fix bootstrap loop | Once `buxc2 == buxc3`, we are self-hosted. Freeze Nim version as reference. |
+**Pre-requisites (Phase 7.0):** StringMap, String split/join, String formatting, File I/O must be working.
+
+| Task | Details | Deps |
+|------|---------|------|
+| `7.1` Port foundation | `token.bux`, `source_location.bux`, `types.bux`, `scope.bux`, `hir.bux` (~600 LOC) | StringMap |
+| `7.2` Port lexer | `lexer.bux` — state machine, UTF-8, error reporting (~570 LOC) | String split/compare |
+| `7.3` Port AST + parser | `ast.bux` + `parser.bux` — Pratt parser, algebraic enums (~1620 LOC) | Array<T>, match |
+| `7.4` Port sema | `sema.bux` — type checking, symbol resolution (~890 LOC) | StringMap, formatting |
+| `7.5` Port manifest | `manifest.bux` — TOML/bux.toml parser (~80 LOC) | File I/O, String split |
+| `7.6` Port HIR lowering | `hir_lower.bux` — tree transformation (~1230 LOC) | StringMap, HashSet |
+| `7.7` Port C backend | `c_backend.bux` — C code generator (~520 LOC) | String formatting |
+| `7.8` Port CLI | `cli.bux` + `main.bux` — command dispatch (~400 LOC) | File I/O, path ops |
+| `7.9` Dogfooding | Use `buxc` (Nim) to build `buxc2` (Bux). Then use `buxc2` to build `buxc3`. Compare bit-for-bit. | All of above |
+| `7.10` Fix bootstrap loop | Once `buxc2 == buxc3`, we are self-hosted. Freeze Nim version as reference. | 7.9 |
 
 **Deliverable:** `make selfhost` succeeds; Bux compiler is written entirely in Bux.
 
@@ -522,13 +621,21 @@ func Main() -> int {
 
 ---
 
-## Next Immediate Steps
+## Next Immediate Steps ✅ (Completed 2026-05-31)
 
-1. **Inferred generic function calls** — `UseBox(&b)` instead of `UseBox<int>(&b)`
-2. **`extend Box<T>` syntax** — Parser support for generic impl blocks
-3. **Std::String improvements** — String builder, slicing, interpolation
-4. **Std::Map generic** — `Map<K,V>` with generic keys and values
-5. **Self-hosting preparation** — Audit Nim compiler code for Bux-rewrite feasibility
+1. ✅ **Inferred generic function calls** — `Max(10, 20)` instead of `Max<int>(10, 20)`
+2. ✅ **`extend Box<T>` syntax** — Parser support for generic impl blocks
+3. ✅ **Std::String improvements** — Slicing, trimming, contains, StringBuilder
+4. ✅ **Std::Map generic** — `Map<K,V>` with generic keys and values (value-type keys)
+5. ✅ **Self-hosting audit** — See Phase 6.5 below
+
+### Next Actions (Priority order)
+
+1. **StringMap** — `Map<String, V>` with strcmp-based key comparison (blocker #1)
+2. **String split/join** — `String_Split(s, sep)`, `String_Join(parts, sep)` for parsing
+3. **String formatting** — Simple `String_Format(pattern, args...)` or `sb.Append(...)` pattern
+4. **File I/O** — `readFile`, `writeFile`, `fileExists` via C stdio
+5. **Begin porting lexer** — Start with `token.bux` + `lexer.bux` (most self-contained modules)
 
 ---
 
diff --git a/docs/LanguageRef.md b/docs/LanguageRef.md
index f1667e6..acb2c7f 100644
--- a/docs/LanguageRef.md
+++ b/docs/LanguageRef.md
@@ -302,23 +302,55 @@ extend Rectangle {
 
 ## Generics
 
-Generic functions are monomorphized at compile time:
+### Generic Functions
+
+Generic functions are monomorphized at compile time. Type parameters can be specified explicitly or inferred from arguments:
 
 ```bux
-func Swap<T>(a: *T, b: *T) {
-    let tmp: T = *a;
-    *a = *b;
-    *b = tmp;
+func Max<T>(a: T, b: T) -> T {
+    if a > b { return a; }
+    return b;
 }
 
 func Main() -> int {
-    let x: int = 1;
-    let y: int = 2;
-    Swap<int>(&x, &y);
+    // Explicit type args
+    let m1: int = Max<int>(10, 20);
+
+    // Type inference — T inferred as int from arguments
+    let m2: int = Max(10, 20);
     return 0;
 }
 ```
 
+### Generic Structs
+
+```bux
+struct Box<T> {
+    value: T,
+}
+
+// Use extend Type<T> for methods on generic structs
+extend Box<T> {
+    func Get(self: *Box<T>) -> T {
+        return self.value;
+    }
+
+    func Set(self: *Box<T>, value: T) {
+        self.value = value;
+    }
+}
+
+func Main() -> int {
+    let b: Box<int> = Box<int> { value: 42 };
+    PrintInt(b.Get());  // 42
+    b.Set(100);
+    PrintInt(b.Get());  // 100
+    return 0;
+}
+```
+
+> **Note:** `extend Type<T>` syntax requires type parameters on the impl block. The compiler propagates them to each method automatically.
+
 ---
 
 ## Error Handling
diff --git a/docs/Stdlib.md b/docs/Stdlib.md
index 1199b95..f375a36 100644
--- a/docs/Stdlib.md
+++ b/docs/Stdlib.md
@@ -1,12 +1,12 @@
 # Bux Standard Library
 
-The Bux standard library provides core functionality for systems programming. It is intentionally minimal for the bootstrap phase and will grow as the language matures.
+The Bux standard library provides core functionality for systems programming. All modules are merged into every compilation, so no explicit linking is needed.
 
 ---
 
 ## Std::Io
 
-Basic input/output operations wrapping C stdio.
+Basic I/O and file operations wrapping C stdio.
 
 ### Functions
 
@@ -18,14 +18,17 @@ Basic input/output operations wrapping C stdio.
 | `PrintFloat` | `func PrintFloat(f: float64)` | Print float |
 | `PrintBool` | `func PrintBool(b: bool)` | Print boolean |
 | `ReadLine` | `func ReadLine() -> String` | Read line from stdin |
+| `ReadFile` | `func ReadFile(path: String) -> String` | Read entire file into string |
+| `WriteFile` | `func WriteFile(path: String, content: String) -> bool` | Write string to file |
+| `FileExists` | `func FileExists(path: String) -> bool` | Check if file exists |
 
 ### Example
 ```bux
-import Std::Io::{PrintLine, PrintInt};
+import Std::Io::{PrintLine, ReadFile, WriteFile};
 
 func Main() -> int {
-    PrintLine("Hello, World!");
-    PrintInt(42);
+    WriteFile("/tmp/test.txt", "Hello, Bux!");
+    PrintLine(ReadFile("/tmp/test.txt"));
     return 0;
 }
 ```
@@ -34,13 +37,12 @@ func Main() -> int {
 
 ## Std::Array
 
-Dynamic array of integers (currently hardcoded for `int`).
+Fully generic dynamic array `Array<T>`.
 
 ### Types
-
 ```bux
-struct Array {
-    data: *int,
+struct Array<T> {
+    data: *T,
     len: uint,
     cap: uint,
 }
@@ -50,22 +52,22 @@ struct Array {
 
 | Function | Signature | Description |
 |----------|-----------|-------------|
-| `Array_New` | `func Array_New(cap: uint) -> Array` | Create new array with capacity |
-| `Array_Push` | `func Array_Push(arr: *Array, value: int)` | Append element |
-| `Array_Get` | `func Array_Get(arr: *Array, index: uint) -> int` | Get element at index |
-| `Array_Len` | `func Array_Len(arr: *Array) -> uint` | Get length |
-| `Array_Free` | `func Array_Free(arr: *Array)` | Free memory |
+| `Array_New<T>` | `func Array_New<T>(cap: uint) -> Array<T>` | Create new array |
+| `Array_Push<T>` | `func Array_Push<T>(arr: *Array<T>, value: T)` | Append element |
+| `Array_Get<T>` | `func Array_Get<T>(arr: *Array<T>, index: uint) -> T` | Get element at index |
+| `Array_Len<T>` | `func Array_Len<T>(arr: *Array<T>) -> uint` | Get length |
+| `Array_Free<T>` | `func Array_Free<T>(arr: *Array<T>)` | Free memory |
 
 ### Example
 ```bux
 import Std::Array::{Array, Array_New, Array_Push, Array_Get};
 
 func Main() -> int {
-    let arr: Array = Array_New(4);
-    Array_Push(&arr, 10);
-    Array_Push(&arr, 20);
-    PrintInt(Array_Get(&arr, 0));  // 10
-    Array_Free(&arr);
+    let arr: Array<int> = Array_New<int>(4);
+    Array_Push<int>(&arr, 10);
+    Array_Push<int>(&arr, 20);
+    PrintInt(Array_Get<int>(&arr, 0));  // 10
+    Array_Free<int>(&arr);
     return 0;
 }
 ```
@@ -74,28 +76,81 @@ func Main() -> int {
 
 ## Std::String
 
-String manipulation utilities for the built-in `String` type (`const char*` in C).
+String manipulation utilities.
+
+### Core Functions
+
+| Function | Signature | Description |
+|----------|-----------|-------------|
+| `String_Len` | `func String_Len(s: String) -> uint` | Length of string |
+| `String_Eq` | `func String_Eq(a: String, b: String) -> bool` | Compare strings |
+| `String_Concat` | `func String_Concat(a: String, b: String) -> String` | Concatenate (allocates) |
+| `String_Copy` | `func String_Copy(s: String) -> String` | Copy string (allocates) |
+| `String_StartsWith` | `func String_StartsWith(s: String, prefix: String) -> bool` | Check prefix |
+| `String_EndsWith` | `func String_EndsWith(s: String, suffix: String) -> bool` | Check suffix |
+| `String_Contains` | `func String_Contains(s: String, substr: String) -> bool` | Check substring |
+
+### Slicing & Trimming
+
+| Function | Signature | Description |
+|----------|-----------|-------------|
+| `String_Slice` | `func String_Slice(s: String, start: uint, len: uint) -> String` | Extract substring |
+| `String_Trim` | `func String_Trim(s: String) -> String` | Trim both sides |
+| `String_TrimLeft` | `func String_TrimLeft(s: String) -> String` | Trim left side |
+| `String_TrimRight` | `func String_TrimRight(s: String) -> String` | Trim right side |
+
+### Split & Join
+
+| Function | Signature | Description |
+|----------|-----------|-------------|
+| `String_SplitCount` | `func String_SplitCount(s: String, delim: String) -> uint` | Count parts |
+| `String_SplitPart` | `func String_SplitPart(s: String, delim: String, index: uint) -> String` | Get nth part (0-indexed) |
+| `String_Join2` | `func String_Join2(a: String, b: String, sep: String) -> String` | Join two strings with separator |
+
+### Conversion
+
+| Function | Signature | Description |
+|----------|-----------|-------------|
+| `String_FromInt` | `func String_FromInt(n: int64) -> String` | Int to string |
+| `String_ToInt` | `func String_ToInt(s: String) -> int64` | String to int |
+
+---
+
+## Std::StringBuilder
+
+Efficient string construction using a dynamic buffer.
+
+### Types
+```bux
+struct StringBuilder {
+    handle: *void,
+}
+```
 
 ### Functions
 
 | Function | Signature | Description |
 |----------|-----------|-------------|
-| `String_Len` | `func String_Len(s: String) -> uint` | Length of string (wraps `strlen`) |
-| `String_Eq` | `func String_Eq(a: String, b: String) -> bool` | Compare strings for equality |
-| `String_Concat` | `func String_Concat(a: String, b: String) -> String` | Concatenate two strings (allocates) |
-| `String_Copy` | `func String_Copy(s: String) -> String` | Copy string (allocates) |
-| `String_StartsWith` | `func String_StartsWith(s: String, prefix: String) -> bool` | Check prefix |
+| `StringBuilder_New` | `func StringBuilder_New() -> StringBuilder` | Create with default capacity (64) |
+| `StringBuilder_NewCap` | `func StringBuilder_NewCap(cap: uint) -> StringBuilder` | Create with custom capacity |
+| `StringBuilder_Append` | `func StringBuilder_Append(sb: *StringBuilder, s: String)` | Append string |
+| `StringBuilder_AppendInt` | `func StringBuilder_AppendInt(sb: *StringBuilder, n: int64)` | Append integer |
+| `StringBuilder_AppendFloat` | `func StringBuilder_AppendFloat(sb: *StringBuilder, f: float64)` | Append float |
+| `StringBuilder_AppendChar` | `func StringBuilder_AppendChar(sb: *StringBuilder, c: char8)` | Append char |
+| `StringBuilder_Build` | `func StringBuilder_Build(sb: *StringBuilder) -> String` | Build result string |
+| `StringBuilder_Free` | `func StringBuilder_Free(sb: *StringBuilder)` | Free memory |
 
 ### Example
 ```bux
-import Std::String::{String_Len, String_Concat, String_Eq};
+import Std::String::{StringBuilder, StringBuilder_New, StringBuilder_Append, StringBuilder_AppendInt, StringBuilder_Build, StringBuilder_Free};
 
 func Main() -> int {
-    let hello: String = "Hello";
-    let world: String = "World";
-    let greeting: String = String_Concat(hello, ", ");
-    let full: String = String_Concat(greeting, world);
-    PrintLine(full);  // "Hello, World"
+    let sb: StringBuilder = StringBuilder_New();
+    StringBuilder_Append(&sb, "Hello, ");
+    StringBuilder_Append(&sb, "World! #");
+    StringBuilder_AppendInt(&sb, 42);
+    PrintLine(StringBuilder_Build(&sb));  // "Hello, World! #42"
+    StringBuilder_Free(&sb);
     return 0;
 }
 ```
@@ -104,19 +159,18 @@ func Main() -> int {
 
 ## Std::Map
 
-Hash map with `String` keys and `int` values using linear probing.
+Generic hash map `Map<K, V>` for value-type keys (int, float, etc.).
 
 ### Types
-
 ```bux
-struct MapEntry {
-    key: String,
-    value: int,
+struct MapEntry<K, V> {
+    key: K,
+    value: V,
     occupied: bool,
 }
 
-struct Map {
-    entries: *MapEntry,
+struct Map<K, V> {
+    entries: *MapEntry<K, V>,
     cap: uint,
     len: uint,
 }
@@ -126,24 +180,81 @@ struct Map {
 
 | Function | Signature | Description |
 |----------|-----------|-------------|
-| `Map_New` | `func Map_New(cap: uint) -> Map` | Create new map with given capacity |
-| `Map_Set` | `func Map_Set(m: *Map, key: String, value: int)` | Insert or update key |
-| `Map_Get` | `func Map_Get(m: *Map, key: String) -> int` | Get value by key (0 if missing) |
-| `Map_Has` | `func Map_Has(m: *Map, key: String) -> bool` | Check if key exists |
-| `Map_Len` | `func Map_Len(m: *Map) -> uint` | Number of entries |
-| `Map_Free` | `func Map_Free(m: *Map)` | Free memory |
+| `Map_New<K,V>` | `func Map_New<K,V>(cap: uint) -> Map<K,V>` | Create map |
+| `Map_Set<K,V>` | `func Map_Set<K,V>(m: *Map<K,V>, key: K, value: V)` | Insert/update |
+| `Map_Get<K,V>` | `func Map_Get<K,V>(m: *Map<K,V>, key: K) -> V` | Get value (zero if missing) |
+| `Map_Has<K,V>` | `func Map_Has<K,V>(m: *Map<K,V>, key: K) -> bool` | Check key exists |
+| `Map_Len<K,V>` | `func Map_Len<K,V>(m: *Map<K,V>) -> uint` | Entry count |
+| `Map_Free<K,V>` | `func Map_Free<K,V>(m: *Map<K,V>)` | Free memory |
 
 ### Example
 ```bux
-import Std::Map::{Map, Map_New, Map_Set, Map_Get, Map_Has};
+import Std::Map::{Map, Map_New, Map_Set, Map_Get, Map_Free};
 
 func Main() -> int {
-    let m: Map = Map_New(16);
-    Map_Set(&m, "one", 1);
-    Map_Set(&m, "two", 2);
-    PrintInt(Map_Get(&m, "one"));   // 1
-    PrintInt(Map_Get(&m, "three")); // 0
-    Map_Free(&m);
+    let m: Map<int, String> = Map_New<int, String>(16);
+    Map_Set<int, String>(&m, 1, "one");
+    Map_Set<int, String>(&m, 2, "two");
+    PrintLine(Map_Get<int, String>(&m, 1));  // "one"
+    Map_Free<int, String>(&m);
+    return 0;
+}
+```
+
+> **Note:** For `String` keys, use `StringMap<V>` below which uses `strcmp` for key comparison.
+
+---
+
+## Std::StringMap
+
+Specialized hash map for `String` keys with any value type.
+
+### Types
+```bux
+struct StringMapEntry<V> {
+    key: String,
+    value: V,
+    occupied: bool,
+}
+
+struct StringMap<V> {
+    entries: *StringMapEntry<V>,
+    cap: uint,
+    len: uint,
+}
+```
+
+### Functions
+
+| Function | Signature | Description |
+|----------|-----------|-------------|
+| `StringMap_New<V>` | `func StringMap_New<V>(cap: uint) -> StringMap<V>` | Create map |
+| `StringMap_Set<V>` | `func StringMap_Set<V>(m: *StringMap<V>, key: String, value: V)` | Insert/update |
+| `StringMap_Get<V>` | `func StringMap_Get<V>(m: *StringMap<V>, key: String) -> V` | Get value |
+| `StringMap_Has<V>` | `func StringMap_Has<V>(m: *StringMap<V>, key: String) -> bool` | Check key exists |
+| `StringMap_Len<V>` | `func StringMap_Len<V>(m: *StringMap<V>) -> uint` | Entry count |
+| `StringMap_Free<V>` | `func StringMap_Free<V>(m: *StringMap<V>)` | Free memory |
+
+---
+
+## Std::Path
+
+Path manipulation utilities.
+
+| Function | Signature | Description |
+|----------|-----------|-------------|
+| `Path_Join` | `func Path_Join(a: String, b: String) -> String` | Join path segments |
+| `Path_Parent` | `func Path_Parent(path: String) -> String` | Get parent directory |
+| `Path_Ext` | `func Path_Ext(path: String) -> String` | Get file extension |
+
+### Example
+```bux
+import Std::Path::{Path_Join, Path_Parent, Path_Ext};
+
+func Main() -> int {
+    PrintLine(Path_Join("/home", "docs"));       // "/home/docs"
+    PrintLine(Path_Parent("/a/b/c.txt"));         // "/a/b"
+    PrintLine(Path_Ext("main.bux"));              // ".bux"
     return 0;
 }
 ```
@@ -156,21 +267,21 @@ These C functions are provided by `runtime.c` and are available via `extern` dec
 
 | Function | Signature | Description |
 |----------|-----------|-------------|
-| `bux_alloc` | `func bux_alloc(size: uint) -> *void` | Allocate memory (wraps `malloc`) |
+| `bux_alloc` | `func bux_alloc(size: uint) -> *void` | Allocate memory |
 | `bux_realloc` | `func bux_realloc(ptr: *void, size: uint) -> *void` | Reallocate memory |
 | `bux_free` | `func bux_free(ptr: *void)` | Free memory |
-| `bux_bounds_check` | `func bux_bounds_check(index: uint, len: uint)` | Panic on out-of-bounds |
+| `bux_bounds_check` | `func bux_bounds_check(index: uint, len: uint)` | Panic on OOB |
+| `bux_hash_string` | `func bux_hash_string(s: String) -> uint` | DJB2 hash for strings |
+| `bux_hash_bytes` | `func bux_hash_bytes(ptr: *void, size: uint) -> uint` | DJB2 hash for raw bytes |
 
 ---
 
 ## Future Modules
 
-Planned for future phases:
-
-- `Std::Result` — Generic `Result<T, E>` with `?` operator support
-- `Std::Option` — Generic `Option<T>`
+- `Std::Result` — Shipped via algebraic enums ✅
+- `Std::Option` — Shipped via algebraic enums ✅
 - `Std::Math` — `Sqrt`, `Pow`, `Min`, `Max`, `Abs`
 - `Std::Os` — `Args`, `Env`, `Exit`, `Cwd`
 - `Std::Fmt` — String formatting with interpolation
 - `Std::Iter` — Iterator trait and combinators
-- `Std::Task` / `Std::Channel` — Lightweight concurrency
+- `Std::Task` / `Std::Channel` — Lightweight concurrency
\ No newline at end of file
diff --git a/examples/extend_generic.bux b/examples/extend_generic.bux
new file mode 100644
index 0000000..5a636ba
--- /dev/null
+++ b/examples/extend_generic.bux
@@ -0,0 +1,30 @@
+// Generic impl blocks — extend Box<T> with methods
+import Std::Io::{PrintLine, PrintInt};
+
+struct Box<T> {
+    value: T,
+}
+
+extend Box<T> {
+    func Get(self: *Box<T>) -> T {
+        return self.value;
+    }
+
+    func Set(self: *Box<T>, value: T) {
+        self.value = value;
+    }
+}
+
+func Main() -> int {
+    let b: Box<int> = Box<int> { value: 42 };
+    PrintLine("Box value:");
+    PrintInt(b.Get());
+    PrintLine("");
+
+    b.Set(100);
+    PrintLine("Box after Set(100):");
+    PrintInt(b.Get());
+    PrintLine("");
+
+    return 0;
+}
\ No newline at end of file
diff --git a/examples/generic_infer.bux b/examples/generic_infer.bux
new file mode 100644
index 0000000..3a1dee0
--- /dev/null
+++ b/examples/generic_infer.bux
@@ -0,0 +1,26 @@
+// Generic function type inference — calls without explicit type args
+import Std::Io::{PrintLine, PrintInt};
+
+func Max<T>(a: T, b: T) -> T {
+    if a > b {
+        return a;
+    } else {
+        return b;
+    }
+}
+
+func Main() -> int {
+    // Inferred: T = int
+    let m1: int = Max(10, 20);
+    let m2: int = Max(5, 3);
+    
+    PrintLine("Max(10, 20) = ");
+    PrintInt(m1);
+    PrintLine("");
+    
+    PrintLine("Max(5, 3) = ");
+    PrintInt(m2);
+    PrintLine("");
+    
+    return 0;
+}
\ No newline at end of file
diff --git a/examples/generic_infer2.bux b/examples/generic_infer2.bux
new file mode 100644
index 0000000..272ae4c
--- /dev/null
+++ b/examples/generic_infer2.bux
@@ -0,0 +1,23 @@
+// Generic function type inference — multiple type parameters
+import Std::Io::{PrintLine, PrintInt};
+
+struct Pair<T, U> {
+    first: T,
+    second: U,
+}
+
+func MakePair<T, U>(a: T, b: U) -> Pair<T, U> {
+    return Pair<T, U> { first: a, second: b };
+}
+
+func Pair_GetFirst<T, U>(self: *Pair<T, U>) -> T {
+    return self.first;
+}
+
+func Main() -> int {
+    // Inferred: T = int, U = String
+    let p: Pair<int, String> = MakePair(42, "hello");
+    PrintInt(p.GetFirst());
+    PrintLine("");
+    return 0;
+}
\ No newline at end of file
diff --git a/examples/map.bux b/examples/map.bux
index 77a629d..2e13e9b 100644
--- a/examples/map.bux
+++ b/examples/map.bux
@@ -1,44 +1,30 @@
-// Map - Hash map with String keys and int values
+// Generic Map<K, V> — uses Std::Map (stdlib is merged into module)
 import Std::Io::{PrintLine, PrintInt};
-import Std::Map::{Map, Map_New, Map_Set, Map_Get, Map_Has, Map_Len};
+import Std::Map::{Map, Map_New, Map_Set, Map_Get, Map_Has, Map_Free};
 
 func Main() -> int {
-    let m: Map = Map_New(16);
+    PrintLine("=== Map<int, String> ===");
+    let m: Map<int, String> = Map_New<int, String>(16);
     
-    PrintLine("Map operations:");
+    Map_Set<int, String>(&m, 1, "one");
+    Map_Set<int, String>(&m, 2, "two");
+    Map_Set<int, String>(&m, 3, "three");
     
-    Map_Set(&m, "one", 1);
-    Map_Set(&m, "two", 2);
-    Map_Set(&m, "three", 3);
+    PrintLine("Get 1:");
+    PrintLine(Map_Get<int, String>(&m, 1));
     
-    PrintLine("Length:");
-    PrintInt(Map_Len(&m));
-    PrintLine("");
+    PrintLine("Get 2:");
+    PrintLine(Map_Get<int, String>(&m, 2));
     
-    PrintLine("Get 'one':");
-    PrintInt(Map_Get(&m, "one"));
-    PrintLine("");
-    
-    PrintLine("Get 'two':");
-    PrintInt(Map_Get(&m, "two"));
-    PrintLine("");
-    
-    PrintLine("Get 'three':");
-    PrintInt(Map_Get(&m, "three"));
-    PrintLine("");
-    
-    Map_Set(&m, "two", 22);
-    PrintLine("Updated 'two':");
-    PrintInt(Map_Get(&m, "two"));
-    PrintLine("");
-    
-    if Map_Has(&m, "one") {
-        PrintLine("Has 'one': yes");
+    if Map_Has<int, String>(&m, 1) {
+        PrintLine("Has 1: yes");
     }
     
-    if !Map_Has(&m, "four") {
-        PrintLine("Has 'four': no");
-    }
+    // Test update
+    Map_Set<int, String>(&m, 2, "updated");
+    PrintLine("Updated 2:");
+    PrintLine(Map_Get<int, String>(&m, 2));
     
+    Map_Free<int, String>(&m);
     return 0;
-}
+}
\ No newline at end of file
diff --git a/examples/strings2.bux b/examples/strings2.bux
new file mode 100644
index 0000000..2e1ea37
--- /dev/null
+++ b/examples/strings2.bux
@@ -0,0 +1,74 @@
+// String improvements — slicing, trimming, contains, StringBuilder
+import Std::Io::{Print, PrintLine, PrintInt, PrintBool};
+import Std::String::{String_Len, String_Slice, String_Contains, String_EndsWith, String_StartsWith};
+import Std::String::{String_Trim, String_TrimLeft, String_TrimRight, String_FromInt, String_ToInt};
+import Std::String::{StringBuilder, StringBuilder_New, StringBuilder_Append, StringBuilder_AppendInt, StringBuilder_Build, StringBuilder_Free};
+
+func Main() -> int {
+    // --- Slicing ---
+    let hello: String = "Hello, World!";
+    Print("Slice(0,5): ");
+    PrintLine(String_Slice(hello, 0, 5));
+
+    Print("Slice(7,5): ");
+    PrintLine(String_Slice(hello, 7, 5));
+
+    // --- Contains ---
+    Print("Contains 'World': ");
+    PrintBool(String_Contains(hello, "World"));
+    PrintLine("");
+
+    Print("Contains 'xyz': ");
+    PrintBool(String_Contains(hello, "xyz"));
+    PrintLine("");
+
+    // --- EndsWith ---
+    Print("EndsWith '!': ");
+    PrintBool(String_EndsWith(hello, "!"));
+    PrintLine("");
+
+    // --- StartsWith ---
+    Print("StartsWith 'Hello': ");
+    PrintBool(String_StartsWith(hello, "Hello"));
+    PrintLine("");
+
+    // --- Trim ---
+    let spaced: String = "   foo bar   ";
+    Print("Trim: '");
+    Print(String_Trim(spaced));
+    PrintLine("'");
+
+    Print("TrimLeft: '");
+    Print(String_TrimLeft(spaced));
+    PrintLine("'");
+
+    Print("TrimRight: '");
+    Print(String_TrimRight(spaced));
+    PrintLine("'");
+
+    // --- Int conversion ---
+    Print("FromInt(42): ");
+    PrintLine(String_FromInt(42));
+
+    Print("FromInt(-100): ");
+    PrintLine(String_FromInt(-100));
+
+    let n: int64 = String_ToInt("123");
+    Print("ToInt('123'): ");
+    PrintInt(n as int);
+    PrintLine("");
+
+    // --- StringBuilder ---
+    let sb: StringBuilder = StringBuilder_New();
+    StringBuilder_Append(&sb, "Hello");
+    StringBuilder_Append(&sb, ", ");
+    StringBuilder_Append(&sb, "World");
+    StringBuilder_Append(&sb, "! #");
+    StringBuilder_AppendInt(&sb, 42);
+    
+    Print("StringBuilder: ");
+    PrintLine(StringBuilder_Build(&sb));
+    StringBuilder_Free(&sb);
+
+    return 0;
+}
\ No newline at end of file
diff --git a/src/ast.nim b/src/ast.nim
index 2305e3d..a10f711 100644
--- a/src/ast.nim
+++ b/src/ast.nim
@@ -162,6 +162,7 @@ type
     of ekCall:
       exprCallCallee*: Expr
       exprCallArgs*: seq[Expr]
+      exprCallInferredTypeArgs*: seq[TypeExpr]  ## filled by sema for inferred generic calls
     of ekGenericCall:
       exprGenericCallee*: string
       exprGenericTypeArgs*: seq[TypeExpr]
@@ -337,6 +338,7 @@ type
       declInterfaceMethods*: seq[Decl]  ## FuncDecl signatures only
     of dkImpl:
       declImplTypeName*: string
+      declImplTypeParams*: seq[string]  ## type parameters for generic impl: extend Box<T>
       declImplInterface*: string        ## empty if not for interface
       declImplMethods*: seq[Decl]
     of dkModule:
diff --git a/src/hir_lower.nim b/src/hir_lower.nim
index 2ace5fc..73175fd 100644
--- a/src/hir_lower.nim
+++ b/src/hir_lower.nim
@@ -508,6 +508,32 @@ proc lowerExpr(ctx: var LowerCtx, expr: Expr): HirNode =
         args.add(ctx.lowerExpr(arg))
       return hirCall(mangledName, args, typ, loc)
 
+    # Inferred generic function call: Max(10, 20) → Max_int(10, 20)
+    if expr.exprCallInferredTypeArgs.len > 0:
+      var calleeName = ""
+      case expr.exprCallCallee.kind
+      of ekIdent:
+        calleeName = expr.exprCallCallee.exprIdent
+        if ctx.importTable.hasKey(calleeName):
+          calleeName = ctx.importTable[calleeName]
+      of ekPath:
+        calleeName = expr.exprCallCallee.exprPath.join("_")
+      else: discard
+      if calleeName != "":
+        var typeSuffix = ""
+        for i, targ in expr.exprCallInferredTypeArgs:
+          if i > 0:
+            typeSuffix.add("_")
+          if targ.kind == tekNamed:
+            typeSuffix.add(targ.typeName)
+          else:
+            typeSuffix.add("unknown")
+        let mangledName = calleeName & "_" & typeSuffix
+        var args: seq[HirNode] = @[]
+        for arg in expr.exprCallArgs:
+          args.add(ctx.lowerExpr(arg))
+        return hirCall(mangledName, args, typ, loc)
+
     # Regular function call
     var calleeName = ""
     if expr.exprCallCallee.kind == ekIdent:
@@ -1004,6 +1030,12 @@ proc lowerModule*(module: Module, sema: Sema): HirModule =
       ctx.genericFuncs[decl.declFuncName] = decl
     if decl.kind == dkStruct and decl.declStructTypeParams.len > 0:
       ctx.genericStructs[decl.declStructName] = decl
+    if decl.kind == dkImpl and decl.declImplTypeParams.len > 0:
+      let typeName = decl.declImplTypeName
+      for methodDecl in decl.declImplMethods:
+        if methodDecl.kind == dkFunc:
+          let mangledName = typeName & "_" & methodDecl.declFuncName
+          ctx.genericFuncs[mangledName] = methodDecl
 
   # Second pass: find all generic calls and monomorphize
   proc findGenericCalls(expr: Expr): seq[tuple[name: string, typeArgs: seq[TypeExpr]]] =
@@ -1013,6 +1045,14 @@ proc lowerModule*(module: Module, sema: Sema): HirModule =
     of ekCall:
       if expr.exprCallCallee.kind == ekGenericCall:
         result.add((expr.exprCallCallee.exprGenericCallee, expr.exprCallCallee.exprGenericTypeArgs))
+      elif expr.exprCallInferredTypeArgs.len > 0:
+        var calleeName = ""
+        case expr.exprCallCallee.kind
+        of ekIdent: calleeName = expr.exprCallCallee.exprIdent
+        of ekPath: calleeName = expr.exprCallCallee.exprPath.join("::")
+        else: discard
+        if calleeName != "":
+          result.add((calleeName, expr.exprCallInferredTypeArgs))
       result.add(findGenericCalls(expr.exprCallCallee))
       for arg in expr.exprCallArgs:
         result.add(findGenericCalls(arg))
@@ -1126,6 +1166,9 @@ proc lowerModule*(module: Module, sema: Sema): HirModule =
     of dkImpl:
       for methodDecl in decl.declImplMethods:
         if methodDecl.kind == dkFunc:
+          # Skip generic methods — they are monomorphized via generateMethodInstance
+          if methodDecl.declFuncTypeParams.len > 0:
+            continue
           var hf = ctx.lowerFunc(methodDecl)
           hf.name = decl.declImplTypeName & "_" & hf.name
           funcs.add(hf)
diff --git a/src/parser.nim b/src/parser.nim
index 7389120..f9d1d9d 100644
--- a/src/parser.nim
+++ b/src/parser.nim
@@ -1015,6 +1015,7 @@ proc parseImplDecl(p: var Parser): Decl =
   let loc = p.currentLoc
   discard p.expect(tkExtend, "expected 'extend'")
   let typeName = p.expect(tkIdent, "expected type name").text
+  let typeParams = p.parseTypeParams()
   var interfaceName = ""
   if p.check(tkFor):
     discard p.advance()
@@ -1030,6 +1031,7 @@ proc parseImplDecl(p: var Parser): Decl =
     methods.add(p.parseFuncDecl(false, false, ParsedAttrs()))
   discard p.expect(tkRBrace, "expected '}' to close impl block")
   return Decl(kind: dkImpl, loc: loc, declImplTypeName: typeName,
+              declImplTypeParams: typeParams,
               declImplInterface: interfaceName, declImplMethods: methods)
 
 proc parseModuleDecl(p: var Parser, isPublic: bool): Decl =
diff --git a/src/sema.nim b/src/sema.nim
index 5e84a79..e2b9ad8 100644
--- a/src/sema.nim
+++ b/src/sema.nim
@@ -47,6 +47,89 @@ proc hasErrors*(res: SemaResult): bool =
       return true
   return false
 
+# ---------------------------------------------------------------------------
+# Generic type inference helpers
+# ---------------------------------------------------------------------------
+
+proc typeExprReferencesTypeParam(te: TypeExpr, name: string): bool =
+  ## Recursively check if a TypeExpr tree references a given type parameter name.
+  if te == nil: return false
+  case te.kind
+  of tekNamed:
+    if te.typeName == name: return true
+    for arg in te.typeArgs:
+      if typeExprReferencesTypeParam(arg, name): return true
+  of tekPath:
+    return false
+  of tekSlice:
+    return typeExprReferencesTypeParam(te.sliceElement, name)
+  of tekPointer:
+    return typeExprReferencesTypeParam(te.pointerPointee, name)
+  of tekTuple:
+    for elem in te.tupleElements:
+      if typeExprReferencesTypeParam(elem, name): return true
+  of tekSelf:
+    return false
+
+proc typeToTypeExpr(t: Type): TypeExpr =
+  ## Convert a resolved Type back to a TypeExpr for storage in inferred type args.
+  case t.kind
+  of tkInt: TypeExpr(kind: tekNamed, typeName: "int")
+  of tkInt8: TypeExpr(kind: tekNamed, typeName: "int8")
+  of tkInt16: TypeExpr(kind: tekNamed, typeName: "int16")
+  of tkInt32: TypeExpr(kind: tekNamed, typeName: "int32")
+  of tkInt64: TypeExpr(kind: tekNamed, typeName: "int64")
+  of tkUInt: TypeExpr(kind: tekNamed, typeName: "uint")
+  of tkUInt8: TypeExpr(kind: tekNamed, typeName: "uint8")
+  of tkUInt16: TypeExpr(kind: tekNamed, typeName: "uint16")
+  of tkUInt32: TypeExpr(kind: tekNamed, typeName: "uint32")
+  of tkUInt64: TypeExpr(kind: tekNamed, typeName: "uint64")
+  of tkFloat32: TypeExpr(kind: tekNamed, typeName: "float32")
+  of tkFloat64: TypeExpr(kind: tekNamed, typeName: "float64")
+  of tkBool: TypeExpr(kind: tekNamed, typeName: "bool")
+  of tkStr: TypeExpr(kind: tekNamed, typeName: "String")
+  of tkNamed: TypeExpr(kind: tekNamed, typeName: t.name)
+  of tkPointer:
+    if t.inner.len > 0:
+      TypeExpr(kind: tekPointer, pointerPointee: typeToTypeExpr(t.inner[0]))
+    else:
+      TypeExpr(kind: tekNamed, typeName: "void")
+  of tkVoid: TypeExpr(kind: tekNamed, typeName: "void")
+  else: TypeExpr(kind: tekNamed, typeName: t.toString)
+
+proc inferTypeArgs(sema: var Sema, funcDecl: Decl, argTypes: seq[Type],
+                   loc: SourceLocation): seq[TypeExpr] =
+  ## Infer type arguments from argument types for a generic function call.
+  ## Returns empty seq if inference fails for any type parameter.
+  result = @[]
+  for tpName in funcDecl.declFuncTypeParams:
+    var inferred: Type = nil
+    for i, param in funcDecl.declFuncParams:
+      if i >= argTypes.len: break
+      # Skip pointer params — type param is inside the pointee and we cannot
+      # structurally extract it (e.g., *Map<K,V> → arg is *Map<int,String>)
+      if param.ptype.kind == tekPointer:
+        continue
+      if typeExprReferencesTypeParam(param.ptype, tpName):
+        if inferred == nil:
+          inferred = argTypes[i]
+        elif inferred != argTypes[i]:
+          # Check if one is assignable to the other (wider type wins)
+          if argTypes[i].isAssignableTo(inferred):
+            discard  # inferred stays the same
+          elif inferred.isAssignableTo(argTypes[i]):
+            inferred = argTypes[i]
+          else:
+            sema.emitError(loc,
+              &"conflicting types for type parameter '{tpName}': " &
+              &"{inferred.toString} vs {argTypes[i].toString}")
+            return @[]
+    if inferred != nil and not inferred.isUnknown:
+      result.add(typeToTypeExpr(inferred))
+    else:
+      # Cannot infer this type parameter from arguments
+      return @[]
+
 # ---------------------------------------------------------------------------
 # Type resolution from AST TypeExpr
 # ---------------------------------------------------------------------------
@@ -229,10 +312,19 @@ proc collectGlobals*(sema: var Sema) =
     of dkImpl:
       # Register methods for the type
       let typeName = decl.declImplTypeName
+      let implTypeParams = decl.declImplTypeParams
       if not sema.methodTable.hasKey(typeName):
         sema.methodTable[typeName] = @[]
+      # If impl has type params, temporarily add them to type table
+      var addedTypeParams: seq[string] = @[]
+      for tp in implTypeParams:
+        sema.typeTable[tp] = makeTypeParam(tp)
+        addedTypeParams.add(tp)
       for methodDecl in decl.declImplMethods:
         if methodDecl.kind == dkFunc:
+          # Propagate impl type params to method for HIR lowering
+          if implTypeParams.len > 0:
+            methodDecl.declFuncTypeParams = implTypeParams
           var params: seq[Type] = @[]
           for p in methodDecl.declFuncParams:
             params.add(sema.resolveType(p.ptype))
@@ -252,7 +344,13 @@ proc collectGlobals*(sema: var Sema) =
           let sym = Symbol(kind: skFunc, name: mangledName, decl: methodDecl,
                            isPublic: true)
           sym.typ = makeFunc(params, retType)
+          if implTypeParams.len > 0:
+            # Register as generic function for monomorphization
+            sym.decl = methodDecl
           discard sema.globalScope.define(sym)
+      # Clean up type parameters
+      for tp in addedTypeParams:
+        sema.typeTable.del(tp)
     else:
       discard
 
@@ -485,6 +583,39 @@ proc checkExpr(sema: var Sema, expr: Expr, scope: Scope): Type =
         for i in 0 ..< argTypes.len:
           if not argTypes[i].isAssignableTo(expectedParams[i]) and not (argTypes[i].kind in {TypeKind.tkUnknown, TypeKind.tkNamed, TypeKind.tkTypeParam}):
             sema.emitError(expr.loc, &"argument {i+1}: expected {expectedParams[i].toString}, got {argTypes[i].toString}")
+
+      # Check for inferred generic function call (no explicit type args)
+      var calleeDecl: Decl = nil
+      case expr.exprCallCallee.kind
+      of ekIdent:
+        let sym = scope.lookup(expr.exprCallCallee.exprIdent)
+        if sym != nil: calleeDecl = sym.decl
+      of ekPath:
+        let fullName = expr.exprCallCallee.exprPath.join("::")
+        let sym = scope.lookup(fullName)
+        if sym != nil: calleeDecl = sym.decl
+      else: discard
+
+      if calleeDecl != nil and calleeDecl.kind == dkFunc and
+         calleeDecl.declFuncTypeParams.len > 0 and
+         expr.exprCallInferredTypeArgs.len == 0 and
+         expr.exprCallCallee.kind != ekGenericCall:
+        let inferred = sema.inferTypeArgs(calleeDecl, argTypes, expr.loc)
+        if inferred.len == calleeDecl.declFuncTypeParams.len:
+          expr.exprCallInferredTypeArgs = inferred
+          # Substitute return type using inferred type args
+          if calleeDecl.declFuncReturnType != nil:
+            var added: seq[string] = @[]
+            for i, tp in calleeDecl.declFuncTypeParams:
+              if i < inferred.len:
+                let concrete = sema.resolveType(inferred[i])
+                sema.typeTable[tp] = concrete
+                added.add(tp)
+            let retType = sema.resolveType(calleeDecl.declFuncReturnType)
+            for tp in added:
+              sema.typeTable.del(tp)
+            return retType
+
       return calleeType.inner[^1]
     elif calleeType.kind == tkUnknown:
       return makeUnknown()
@@ -708,6 +839,10 @@ proc checkStmt(sema: var Sema, stmt: Stmt, scope: Scope): Type =
 proc checkFunc(sema: var Sema, decl: Decl) =
   if decl.declFuncBody == nil:
     return
+  # Skip body type-checking for generic functions — their bodies contain
+  # type parameters that cannot be fully resolved until monomorphization.
+  if decl.declFuncTypeParams.len > 0:
+    return
   var funcScope = newScope(sema.globalScope)
   # Add type parameters to type table for resolution
   var addedTypeParams: seq[string] = @[]
diff --git a/stdlib/Std/Io.bux b/stdlib/Std/Io.bux
index c819829..ea0cf02 100644
--- a/stdlib/Std/Io.bux
+++ b/stdlib/Std/Io.bux
@@ -6,5 +6,22 @@ extern func PrintInt(n: int);
 extern func PrintFloat(f: float64);
 extern func PrintBool(b: bool);
 extern func ReadLine() -> String;
+extern func bux_read_file(path: String) -> String;
+extern func bux_write_file(path: String, content: String) -> int;
+extern func bux_file_exists(path: String) -> int;
+
+func ReadFile(path: String) -> String {
+    return bux_read_file(path);
+}
+
+func WriteFile(path: String, content: String) -> bool {
+    let r: int = bux_write_file(path, content);
+    return r != 0;
+}
+
+func FileExists(path: String) -> bool {
+    let r: int = bux_file_exists(path);
+    return r != 0;
+}
 
 }
diff --git a/stdlib/Std/Map.bux b/stdlib/Std/Map.bux
index 59b8b0a..650803e 100644
--- a/stdlib/Std/Map.bux
+++ b/stdlib/Std/Map.bux
@@ -1,31 +1,120 @@
 module Std::Map {
 
+extern func bux_hash_bytes(ptr: *void, size: uint) -> uint;
 extern func bux_hash_string(s: String) -> uint;
 
-struct MapEntry {
-    key: String,
-    value: int,
+// ---------------------------------------------------------------------------
+// Generic Map<K, V> — works with value-type keys (int, float, etc.)
+// For String keys, use StringMap below.
+// ---------------------------------------------------------------------------
+
+struct MapEntry<K, V> {
+    key: K,
+    value: V,
     occupied: bool,
 }
 
-struct Map {
-    entries: *MapEntry,
+struct Map<K, V> {
+    entries: *MapEntry<K, V>,
     cap: uint,
     len: uint,
 }
 
-func Map_New(cap: uint) -> Map {
-    let total: uint = cap * sizeof(MapEntry);
-    let data: *MapEntry = bux_alloc(total) as *MapEntry;
+func Map_New<K, V>(cap: uint) -> Map<K, V> {
+    let total: uint = cap * sizeof(MapEntry<K, V>);
+    let data: *MapEntry<K, V> = bux_alloc(total) as *MapEntry<K, V>;
     var i: uint = 0;
     while i < cap {
         data[i].occupied = false;
         i = i + 1;
     }
-    return Map { entries: data, cap: cap, len: 0 };
+    return Map<K, V> { entries: data, cap: cap, len: 0 };
 }
 
-func Map_Set(m: *Map, key: String, value: int) {
+func Map_Set<K, V>(m: *Map<K, V>, key: K, value: V) {
+    var keyPtr: *K = &key;
+    let hash: uint = bux_hash_bytes(keyPtr as *void, sizeof(K));
+    var idx: uint = hash % m.cap;
+    while m.entries[idx].occupied {
+        if m.entries[idx].key == key {
+            m.entries[idx].value = value;
+            return;
+        }
+        idx = (idx + 1) % m.cap;
+    }
+    m.entries[idx].key = key;
+    m.entries[idx].value = value;
+    m.entries[idx].occupied = true;
+    m.len = m.len + 1;
+}
+
+func Map_Get<K, V>(m: *Map<K, V>, key: K) -> V {
+    var keyPtr: *K = &key;
+    let hash: uint = bux_hash_bytes(keyPtr as *void, sizeof(K));
+    var idx: uint = hash % m.cap;
+    while m.entries[idx].occupied {
+        if m.entries[idx].key == key {
+            return m.entries[idx].value;
+        }
+        idx = (idx + 1) % m.cap;
+    }
+    // Return zero value for missing key
+    var zero: V = 0 as V;
+    return zero;
+}
+
+func Map_Has<K, V>(m: *Map<K, V>, key: K) -> bool {
+    var keyPtr: *K = &key;
+    let hash: uint = bux_hash_bytes(keyPtr as *void, sizeof(K));
+    var idx: uint = hash % m.cap;
+    while m.entries[idx].occupied {
+        if m.entries[idx].key == key {
+            return true;
+        }
+        idx = (idx + 1) % m.cap;
+    }
+    return false;
+}
+
+func Map_Len<K, V>(m: *Map<K, V>) -> uint {
+    return m.len;
+}
+
+func Map_Free<K, V>(m: *Map<K, V>) {
+    bux_free(m.entries as *void);
+    m.entries = null as *MapEntry<K, V>;
+    m.cap = 0;
+    m.len = 0;
+}
+
+// ---------------------------------------------------------------------------
+// StringMap<V> — specialized Map for String keys, using strcmp
+// ---------------------------------------------------------------------------
+
+struct StringMapEntry<V> {
+    key: String,
+    value: V,
+    occupied: bool,
+}
+
+struct StringMap<V> {
+    entries: *StringMapEntry<V>,
+    cap: uint,
+    len: uint,
+}
+
+func StringMap_New<V>(cap: uint) -> StringMap<V> {
+    let total: uint = cap * sizeof(StringMapEntry<V>);
+    let data: *StringMapEntry<V> = bux_alloc(total) as *StringMapEntry<V>;
+    var i: uint = 0;
+    while i < cap {
+        data[i].occupied = false;
+        i = i + 1;
+    }
+    return StringMap<V> { entries: data, cap: cap, len: 0 };
+}
+
+func StringMap_Set<V>(m: *StringMap<V>, key: String, value: V) {
     let hash: uint = bux_hash_string(key);
     var idx: uint = hash % m.cap;
     while m.entries[idx].occupied {
@@ -41,7 +130,7 @@ func Map_Set(m: *Map, key: String, value: int) {
     m.len = m.len + 1;
 }
 
-func Map_Get(m: *Map, key: String) -> int {
+func StringMap_Get<V>(m: *StringMap<V>, key: String) -> V {
     let hash: uint = bux_hash_string(key);
     var idx: uint = hash % m.cap;
     while m.entries[idx].occupied {
@@ -50,10 +139,11 @@ func Map_Get(m: *Map, key: String) -> int {
         }
         idx = (idx + 1) % m.cap;
     }
-    return 0;
+    var zero: V = 0 as V;
+    return zero;
 }
 
-func Map_Has(m: *Map, key: String) -> bool {
+func StringMap_Has<V>(m: *StringMap<V>, key: String) -> bool {
     let hash: uint = bux_hash_string(key);
     var idx: uint = hash % m.cap;
     while m.entries[idx].occupied {
@@ -65,15 +155,15 @@ func Map_Has(m: *Map, key: String) -> bool {
     return false;
 }
 
-func Map_Len(m: *Map) -> uint {
+func StringMap_Len<V>(m: *StringMap<V>) -> uint {
     return m.len;
 }
 
-func Map_Free(m: *Map) {
+func StringMap_Free<V>(m: *StringMap<V>) {
     bux_free(m.entries as *void);
-    m.entries = null as *MapEntry;
+    m.entries = null as *StringMapEntry<V>;
     m.cap = 0;
     m.len = 0;
 }
 
-}
+}
\ No newline at end of file
diff --git a/stdlib/Std/Path.bux b/stdlib/Std/Path.bux
new file mode 100644
index 0000000..b559c30
--- /dev/null
+++ b/stdlib/Std/Path.bux
@@ -0,0 +1,19 @@
+module Std::Path {
+
+extern func bux_path_join(a: String, b: String) -> String;
+extern func bux_path_parent(path: String) -> String;
+extern func bux_path_ext(path: String) -> String;
+
+func Path_Join(a: String, b: String) -> String {
+    return bux_path_join(a, b);
+}
+
+func Path_Parent(path: String) -> String {
+    return bux_path_parent(path);
+}
+
+func Path_Ext(path: String) -> String {
+    return bux_path_ext(path);
+}
+
+}
\ No newline at end of file
diff --git a/stdlib/Std/String.bux b/stdlib/Std/String.bux
index 5b5b063..7012ee9 100644
--- a/stdlib/Std/String.bux
+++ b/stdlib/Std/String.bux
@@ -6,6 +6,25 @@ extern func bux_strncmp(a: String, b: String, n: uint) -> int;
 extern func bux_strcpy(dest: *char8, src: String) -> *char8;
 extern func bux_strcat(dest: *char8, src: String) -> *char8;
 extern func bux_strncpy(dest: *char8, src: String, n: uint) -> *char8;
+extern func bux_strstr(haystack: String, needle: String) -> String;
+extern func bux_str_contains(haystack: String, needle: String) -> int;
+extern func bux_str_slice(s: String, start: uint, len: uint) -> String;
+extern func bux_str_trim_left(s: String) -> String;
+extern func bux_str_trim_right(s: String) -> String;
+extern func bux_str_trim(s: String) -> String;
+extern func bux_int_to_str(n: int64) -> String;
+extern func bux_str_to_int(s: String) -> int64;
+extern func bux_sb_new(initial_cap: uint) -> *void;
+extern func bux_sb_append(sb: *void, s: String);
+extern func bux_sb_append_int(sb: *void, n: int64);
+extern func bux_sb_append_float(sb: *void, f: float64);
+extern func bux_sb_append_char(sb: *void, c: char8);
+extern func bux_sb_build(sb: *void) -> String;
+extern func bux_sb_free(sb: *void);
+extern func bux_str_split_count(s: String, delim: String) -> uint;
+extern func bux_str_split_part(s: String, delim: String, index: uint) -> String;
+extern func bux_str_join2(a: String, b: String, sep: String) -> String;
+
 
 func String_Len(s: String) -> uint {
     return bux_strlen(s);
@@ -42,4 +61,98 @@ func String_StartsWith(s: String, prefix: String) -> bool {
     return r == 0;
 }
 
+func String_EndsWith(s: String, suffix: String) -> bool {
+    let s_len: uint = bux_strlen(s);
+    let suf_len: uint = bux_strlen(suffix);
+    if suf_len > s_len {
+        return false;
+    }
+    let start: uint = s_len - suf_len;
+    let tail: String = bux_str_slice(s, start, suf_len);
+    let eq: bool = bux_strcmp(tail, suffix) == 0;
+    return eq;
 }
+
+func String_Contains(s: String, substr: String) -> bool {
+    let r: int = bux_str_contains(s, substr);
+    return r != 0;
+}
+
+func String_Slice(s: String, start: uint, len: uint) -> String {
+    return bux_str_slice(s, start, len);
+}
+
+func String_Trim(s: String) -> String {
+    return bux_str_trim(s);
+}
+
+func String_TrimLeft(s: String) -> String {
+    return bux_str_trim_left(s);
+}
+
+func String_TrimRight(s: String) -> String {
+    return bux_str_trim_right(s);
+}
+
+func String_FromInt(n: int64) -> String {
+    return bux_int_to_str(n);
+}
+
+func String_ToInt(s: String) -> int64 {
+    return bux_str_to_int(s);
+}
+
+// String Builder — efficient string construction
+struct StringBuilder {
+    handle: *void,
+}
+
+func StringBuilder_New() -> StringBuilder {
+    return StringBuilder { handle: bux_sb_new(64) };
+}
+
+func StringBuilder_NewCap(cap: uint) -> StringBuilder {
+    return StringBuilder { handle: bux_sb_new(cap) };
+}
+
+func StringBuilder_Append(sb: *StringBuilder, s: String) {
+    bux_sb_append(sb.handle, s);
+}
+
+func StringBuilder_AppendInt(sb: *StringBuilder, n: int64) {
+    bux_sb_append_int(sb.handle, n);
+}
+
+func StringBuilder_AppendFloat(sb: *StringBuilder, f: float64) {
+    bux_sb_append_float(sb.handle, f);
+}
+
+func StringBuilder_AppendChar(sb: *StringBuilder, c: char8) {
+    bux_sb_append_char(sb.handle, c);
+}
+
+func StringBuilder_Build(sb: *StringBuilder) -> String {
+    return bux_sb_build(sb.handle);
+}
+
+func StringBuilder_Free(sb: *StringBuilder) {
+    bux_sb_free(sb.handle);
+}
+
+// ---------------------------------------------------------------------------
+// String split/join
+// ---------------------------------------------------------------------------
+
+func String_SplitCount(s: String, delim: String) -> uint {
+    return bux_str_split_count(s, delim);
+}
+
+func String_SplitPart(s: String, delim: String, index: uint) -> String {
+    return bux_str_split_part(s, delim, index);
+}
+
+func String_Join2(a: String, b: String, sep: String) -> String {
+    return bux_str_join2(a, b, sep);
+}
+
+}
\ No newline at end of file
diff --git a/stdlib/runtime.c b/stdlib/runtime.c
index 37a09c5..4cb2a99 100644
--- a/stdlib/runtime.c
+++ b/stdlib/runtime.c
@@ -153,6 +153,391 @@ char* bux_strncpy(char* dest, const char* src, unsigned int n) {
     return strncpy(dest, src, (size_t)n);
 }
 
+/* String find: returns pointer to first occurrence of needle in haystack, or NULL */
+const char* bux_strstr(const char* haystack, const char* needle) {
+    if (!haystack || !needle) return NULL;
+    return strstr(haystack, needle);
+}
+
+/* String contains: returns 1 if haystack contains needle, 0 otherwise */
+int bux_str_contains(const char* haystack, const char* needle) {
+    return bux_strstr(haystack, needle) != NULL;
+}
+
+/* String slice: extract substring from start, length len */
+char* bux_str_slice(const char* s, unsigned int start, unsigned int len) {
+    if (!s) return NULL;
+    unsigned int s_len = (unsigned int)strlen(s);
+    if (start >= s_len) {
+        char* empty = (char*)bux_alloc(1);
+        empty[0] = '\0';
+        return empty;
+    }
+    unsigned int avail = s_len - start;
+    if (len > avail) len = avail;
+    char* result = (char*)bux_alloc(len + 1);
+    memcpy(result, s + start, len);
+    result[len] = '\0';
+    return result;
+}
+
+/* String trim left: remove leading whitespace */
+char* bux_str_trim_left(const char* s) {
+    if (!s) return NULL;
+    while (*s == ' ' || *s == '\t' || *s == '\n' || *s == '\r') s++;
+    unsigned int len = (unsigned int)strlen(s);
+    char* result = (char*)bux_alloc(len + 1);
+    memcpy(result, s, len + 1);
+    return result;
+}
+
+/* String trim right: remove trailing whitespace */
+char* bux_str_trim_right(const char* s) {
+    if (!s) return NULL;
+    unsigned int len = (unsigned int)strlen(s);
+    while (len > 0 && (s[len-1] == ' ' || s[len-1] == '\t' || s[len-1] == '\n' || s[len-1] == '\r')) {
+        len--;
+    }
+    char* result = (char*)bux_alloc(len + 1);
+    memcpy(result, s, len);
+    result[len] = '\0';
+    return result;
+}
+
+/* String trim: remove both leading and trailing whitespace */
+char* bux_str_trim(const char* s) {
+    char* left = bux_str_trim_left(s);
+    char* result = bux_str_trim_right(left);
+    bux_free(left);
+    return result;
+}
+
+/* Int to string conversion */
+char* bux_int_to_str(int64_t n) {
+    char* result = (char*)bux_alloc(32);
+    snprintf(result, 32, "%lld", (long long)n);
+    return result;
+}
+
+/* String to int conversion */
+int64_t bux_str_to_int(const char* s) {
+    if (!s) return 0;
+    return (int64_t)atoll(s);
+}
+
+/* String builder */
+typedef struct {
+    char* buf;
+    unsigned int len;
+    unsigned int cap;
+} BuxStringBuilder;
+
+BuxStringBuilder* bux_sb_new(unsigned int initial_cap) {
+    BuxStringBuilder* sb = (BuxStringBuilder*)bux_alloc(sizeof(BuxStringBuilder));
+    sb->cap = initial_cap > 0 ? initial_cap : 64;
+    sb->buf = (char*)bux_alloc(sb->cap);
+    sb->buf[0] = '\0';
+    sb->len = 0;
+    return sb;
+}
+
+void bux_sb_append(BuxStringBuilder* sb, const char* s) {
+    if (!sb || !s) return;
+    unsigned int s_len = (unsigned int)strlen(s);
+    unsigned int new_len = sb->len + s_len;
+    if (new_len + 1 > sb->cap) {
+        while (sb->cap < new_len + 1) sb->cap *= 2;
+        sb->buf = (char*)bux_realloc(sb->buf, sb->cap);
+    }
+    memcpy(sb->buf + sb->len, s, s_len);
+    sb->len = new_len;
+    sb->buf[sb->len] = '\0';
+}
+
+void bux_sb_append_int(BuxStringBuilder* sb, int64_t n) {
+    char tmp[32];
+    snprintf(tmp, sizeof(tmp), "%lld", (long long)n);
+    bux_sb_append(sb, tmp);
+}
+
+void bux_sb_append_float(BuxStringBuilder* sb, double f) {
+    char tmp[32];
+    snprintf(tmp, sizeof(tmp), "%g", f);
+    bux_sb_append(sb, tmp);
+}
+
+void bux_sb_append_char(BuxStringBuilder* sb, char c) {
+    if (!sb) return;
+    if (sb->len + 2 > sb->cap) {
+        sb->cap *= 2;
+        sb->buf = (char*)bux_realloc(sb->buf, sb->cap);
+    }
+    sb->buf[sb->len++] = c;
+    sb->buf[sb->len] = '\0';
+}
+
+const char* bux_sb_build(BuxStringBuilder* sb) {
+    if (!sb) return "";
+    return sb->buf;
+}
+
+void bux_sb_free(BuxStringBuilder* sb) {
+    if (!sb) return;
+    bux_free(sb->buf);
+    bux_free(sb);
+}
+
+/* String split: count parts separated by delimiter */
+unsigned int bux_str_split_count(const char* s, const char* delim) {
+    if (!s || !delim || !*delim) return 1;
+    unsigned int count = 1;
+    size_t delim_len = strlen(delim);
+    const char* p = s;
+    while ((p = strstr(p, delim)) != NULL) {
+        count++;
+        p += delim_len;
+    }
+    return count;
+}
+
+/* String split: get the n-th part (0-indexed) */
+char* bux_str_split_part(const char* s, const char* delim, unsigned int index) {
+    if (!s || !delim || !*delim) {
+        if (index == 0) {
+            unsigned int len = s ? (unsigned int)strlen(s) : 0;
+            char* result = (char*)bux_alloc(len + 1);
+            if (s) memcpy(result, s, len);
+            result[len] = '\0';
+            return result;
+        }
+        char* empty = (char*)bux_alloc(1);
+        empty[0] = '\0';
+        return empty;
+    }
+    size_t delim_len = strlen(delim);
+    const char* start = s;
+    const char* end;
+    unsigned int current = 0;
+    while (current < index) {
+        end = strstr(start, delim);
+        if (!end) {
+            char* empty = (char*)bux_alloc(1);
+            empty[0] = '\0';
+            return empty;
+        }
+        start = end + delim_len;
+        current++;
+    }
+    end = strstr(start, delim);
+    size_t part_len;
+    if (end) {
+        part_len = (size_t)(end - start);
+    } else {
+        part_len = strlen(start);
+    }
+    char* result = (char*)bux_alloc(part_len + 1);
+    memcpy(result, start, part_len);
+    result[part_len] = '\0';
+    return result;
+}
+
+/* String join: join two strings with separator */
+char* bux_str_join2(const char* a, const char* b, const char* sep) {
+    if (!a && !b) {
+        char* empty = (char*)bux_alloc(1);
+        empty[0] = '\0';
+        return empty;
+    }
+    unsigned int len_a = a ? (unsigned int)strlen(a) : 0;
+    unsigned int len_b = b ? (unsigned int)strlen(b) : 0;
+    unsigned int len_sep = sep ? (unsigned int)strlen(sep) : 0;
+    unsigned int total = len_a + len_sep + len_b;
+    char* result = (char*)bux_alloc(total + 1);
+    if (a) memcpy(result, a, len_a);
+    if (sep && len_a > 0 && len_b > 0) memcpy(result + len_a, sep, len_sep);
+    if (b) memcpy(result + len_a + (len_a > 0 && len_b > 0 ? len_sep : 0), b, len_b);
+    result[total] = '\0';
+    return result;
+}
+
+/* Simple string format: replace {0}, {1}, ... with string arguments.
+   Returns formatted string. Supports up to 8 arguments. */
+char* bux_str_format(const char* pattern,
+    const char* a0, const char* a1, const char* a2, const char* a3,
+    const char* a4, const char* a5, const char* a6, const char* a7) {
+    if (!pattern) {
+        char* empty = (char*)bux_alloc(1);
+        empty[0] = '\0';
+        return empty;
+    }
+    const char* args[8] = { a0, a1, a2, a3, a4, a5, a6, a7 };
+    
+    /* Calculate result size */
+    size_t total = 0;
+    const char* p = pattern;
+    while (*p) {
+        if (*p == '{' && p[1] >= '0' && p[1] <= '7' && p[2] == '}') {
+            int idx = p[1] - '0';
+            if (args[idx]) total += strlen(args[idx]);
+            p += 3;
+        } else {
+            total++;
+            p++;
+        }
+    }
+    
+    char* result = (char*)bux_alloc(total + 1);
+    char* w = result;
+    p = pattern;
+    while (*p) {
+        if (*p == '{' && p[1] >= '0' && p[1] <= '7' && p[2] == '}') {
+            int idx = p[1] - '0';
+            if (args[idx]) {
+                size_t len = strlen(args[idx]);
+                memcpy(w, args[idx], len);
+                w += len;
+            }
+            p += 3;
+        } else {
+            *w++ = *p++;
+        }
+    }
+    *w = '\0';
+    return result;
+}
+
+/* File I/O — read entire file into string */
+char* bux_read_file(const char* path) {
+    if (!path) return NULL;
+    FILE* f = fopen(path, "rb");
+    if (!f) return NULL;
+    fseek(f, 0, SEEK_END);
+    long size = ftell(f);
+    fseek(f, 0, SEEK_SET);
+    char* buf = (char*)bux_alloc((size_t)size + 1);
+    size_t read = fread(buf, 1, (size_t)size, f);
+    fclose(f);
+    buf[read] = '\0';
+    return buf;
+}
+
+/* File I/O — write string to file */
+int bux_write_file(const char* path, const char* content) {
+    if (!path || !content) return 0;
+    FILE* f = fopen(path, "wb");
+    if (!f) return 0;
+    size_t len = strlen(content);
+    size_t written = fwrite(content, 1, len, f);
+    fclose(f);
+    return written == len ? 1 : 0;
+}
+
+/* File I/O — check if file exists */
+int bux_file_exists(const char* path) {
+    if (!path) return 0;
+    FILE* f = fopen(path, "rb");
+    if (f) {
+        fclose(f);
+        return 1;
+    }
+    return 0;
+}
+
+/* Path operations */
+char* bux_path_join(const char* a, const char* b) {
+    if (!a && !b) {
+        char* empty = (char*)bux_alloc(1);
+        empty[0] = '\0';
+        return empty;
+    }
+    if (!a) {
+        unsigned int len = (unsigned int)strlen(b);
+        char* result = (char*)bux_alloc(len + 1);
+        memcpy(result, b, len + 1);
+        return result;
+    }
+    if (!b) {
+        unsigned int len = (unsigned int)strlen(a);
+        char* result = (char*)bux_alloc(len + 1);
+        memcpy(result, a, len + 1);
+        return result;
+    }
+    unsigned int len_a = (unsigned int)strlen(a);
+    unsigned int len_b = (unsigned int)strlen(b);
+    int need_sep = (len_a > 0 && a[len_a-1] != '/') ? 1 : 0;
+    unsigned int total = len_a + (need_sep ? 1 : 0) + len_b;
+    char* result = (char*)bux_alloc(total + 1);
+    memcpy(result, a, len_a);
+    if (need_sep) result[len_a] = '/';
+    memcpy(result + len_a + (need_sep ? 1 : 0), b, len_b);
+    result[total] = '\0';
+    return result;
+}
+
+char* bux_path_parent(const char* path) {
+    if (!path) {
+        char* empty = (char*)bux_alloc(1);
+        empty[0] = '\0';
+        return empty;
+    }
+    int len = (int)strlen(path);
+    while (len > 0 && path[len-1] == '/') len--;
+    while (len > 0 && path[len-1] != '/') len--;
+    while (len > 0 && path[len-1] == '/') len--;
+    if (len == 0) {
+        char* dot = (char*)bux_alloc(2);
+        dot[0] = '.';
+        dot[1] = '\0';
+        return dot;
+    }
+    char* result = (char*)bux_alloc((unsigned int)len + 1);
+    memcpy(result, path, (unsigned int)len);
+    result[len] = '\0';
+    return result;
+}
+
+char* bux_path_ext(const char* path) {
+    if (!path) {
+        char* empty = (char*)bux_alloc(1);
+        empty[0] = '\0';
+        return empty;
+    }
+    const char* dot = strrchr(path, '.');
+    if (!dot) {
+        char* empty = (char*)bux_alloc(1);
+        empty[0] = '\0';
+        return empty;
+    }
+    const char* slash = strrchr(path, '/');
+    if (slash && slash > dot) {
+        char* empty = (char*)bux_alloc(1);
+        empty[0] = '\0';
+        return empty;
+    }
+    unsigned int len = (unsigned int)strlen(dot);
+    char* result = (char*)bux_alloc(len + 1);
+    memcpy(result, dot, len + 1);
+    return result;
+}
+
+/* Hash function (djb2) over raw bytes — for generic key types */
+unsigned int bux_hash_bytes(const void* ptr, size_t size) {
+    if (!ptr) return 0;
+    unsigned int hash = 5381;
+    const unsigned char* bytes = (const unsigned char*)ptr;
+    for (size_t i = 0; i < size; i++) {
+        hash = ((hash << 5) + hash) + bytes[i]; /* hash * 33 + byte */
+    }
+    return hash;
+}
+
+/* Byte equality check — for generic key comparison */
+int bux_mem_eq(const void* a, const void* b, size_t size) {
+    if (a == b) return 1;
+    if (!a || !b) return 0;
+    return memcmp(a, b, size) == 0;
+}
+
 /* Hash function (djb2) for string keys */
 unsigned int bux_hash_string(const char* s) {
     unsigned int hash = 5381;