diff --git a/CHANGELOG.md b/CHANGELOG.md
index 56b6a9a..826eefd 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -16,6 +16,14 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Fixed
 - Upcoming fixes will be listed here
 
+## [0.1.3] - 2025-08-19
+
+### Fixed
+- CLI crash on `--help` due to unsupported `Optional[typer.Context]` annotation. Refactored callback to require `Context` and added an internal wrapper for tests.
+
+### Changed
+- Documentation updates clarifying correct usage of global options and subcommands. Added examples for `nl` subcommand and a note to avoid invoking the binary twice.
+
 ## [0.1.0] - 2024-01-XX
 
 ### Added
diff --git a/README.md b/README.md
index b813b1a..4f1c208 100644
--- a/README.md
+++ b/README.md
@@ -137,6 +137,22 @@ aiclip --timeout 120 "complex processing task"
 aiclip --verbose "your command"
 ```
 
+### Subcommands and option placement
+
+You can also use the explicit `nl` subcommand. Put global options before the subcommand:
+
+```bash
+aiclip --yes nl "thumbnail at 10s from test.mp4"
+aiclip --dry-run --model gpt-4o-mini nl "compress input.mp4"
+```
+
+Do not invoke the binary twice:
+
+```bash
+# Incorrect
+aiclip aiclip --yes nl "..."
+```
+
 ## 🔧 Configuration
 
 aiclip uses environment variables and `.env` files for configuration:
diff --git a/docs/usage.md b/docs/usage.md
index e3bb68c..f3cb9d8 100644
--- a/docs/usage.md
+++ b/docs/usage.md
@@ -10,3 +10,22 @@ Options:
 - `--yes`: skip confirmation
 - `--dry-run`: preview only
 - `--model gpt-4o-mini`: pick model
+
+## Subcommand mode
+
+Global options must come before the subcommand. The primary subcommand is `nl` (natural language):
+
+```bash
+# Run with confirmation skipped
+aiclip --yes nl "thumbnail at 10s from test.mp4"
+
+# Preview only and override model
+aiclip --dry-run --model gpt-4o-mini nl "compress input.mp4"
+```
+
+Avoid invoking the binary twice:
+
+```bash
+# Incorrect
+aiclip aiclip --yes nl "..."
+```
diff --git a/pyproject.toml b/pyproject.toml
index 561f63d..8bda090 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "ai-ffmpeg-cli"
-version = "0.1.2"
+version = "0.1.3"
 description = "AI-powered CLI that translates natural language to safe ffmpeg commands"
 readme = { file = "README.md", content-type = "text/markdown" }
 license = { file = "LICENSE" }
diff --git a/src/ai_ffmpeg_cli/main.py b/src/ai_ffmpeg_cli/main.py
index ae630c5..206c7ab 100644
--- a/src/ai_ffmpeg_cli/main.py
+++ b/src/ai_ffmpeg_cli/main.py
@@ -26,17 +26,14 @@ def _setup_logging(verbose: bool) -> None:
     logging.basicConfig(level=level, format="%(levelname)s: %(message)s")
 
 
-@app.callback()
-def main(
-    ctx: typer.Context | None = None,
-    prompt: str | None = typer.Argument(
-        None, help="Natural language prompt; if provided, runs once and exits"
-    ),
-    yes: bool = typer.Option(False, "--yes/--no-yes", help="Skip confirmation and overwrite"),
-    model: str | None = typer.Option(None, "--model", help="LLM model override"),
-    dry_run: bool = typer.Option(None, "--dry-run/--no-dry-run", help="Preview only"),
-    timeout: int = typer.Option(60, "--timeout", help="LLM timeout seconds"),
-    verbose: bool = typer.Option(False, "--verbose", help="Verbose logging"),
+def _main_impl(
+    ctx: typer.Context | None,
+    prompt: str | None,
+    yes: bool,
+    model: str | None,
+    dry_run: bool | None,
+    timeout: int,
+    verbose: bool,
 ) -> None:
     """Initialize global options and optionally run one-shot prompt."""
     _setup_logging(verbose)
@@ -84,6 +81,33 @@ def main(
         raise typer.Exit(1) from e
 
 
+@app.callback()
+def cli_main(
+    ctx: typer.Context,
+    prompt: str | None = typer.Argument(
+        None, help="Natural language prompt; if provided, runs once and exits"
+    ),
+    yes: bool = typer.Option(False, "--yes/--no-yes", help="Skip confirmation and overwrite"),
+    model: str | None = typer.Option(None, "--model", help="LLM model override"),
+    dry_run: bool = typer.Option(None, "--dry-run/--no-dry-run", help="Preview only"),
+    timeout: int = typer.Option(60, "--timeout", help="LLM timeout seconds"),
+    verbose: bool = typer.Option(False, "--verbose", help="Verbose logging"),
+) -> None:
+    _main_impl(ctx, prompt, yes, model, dry_run, timeout, verbose)
+
+
+def main(
+    ctx: typer.Context | None = None,
+    prompt: str | None = None,
+    yes: bool = False,
+    model: str | None = None,
+    dry_run: bool | None = None,
+    timeout: int = 60,
+    verbose: bool = False,
+) -> None:
+    _main_impl(ctx, prompt, yes, model, dry_run, timeout, verbose)
+
+
 def _make_llm(cfg: AppConfig) -> LLMClient:
     if not cfg.openai_api_key:
         raise ConfigError(