diff --git a/src/ctxlens/cli.py b/src/ctxlens/cli.py
index ef87fdd..3d4835a 100644
--- a/src/ctxlens/cli.py
+++ b/src/ctxlens/cli.py
@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import math
 import sys
 from pathlib import Path
 
@@ -35,6 +36,26 @@
 EXIT_ERROR = 1
 
 
+def _positive_int(value: int) -> int:
+    if value < 1:
+        raise typer.BadParameter("must be a positive integer (>= 1)")
+    return value
+
+
+def _non_negative_int(value: int) -> int:
+    if value < 0:
+        raise typer.BadParameter("must be a non-negative integer (>= 0)")
+    return value
+
+
+def _ratio(value: float | None) -> float | None:
+    if value is None:
+        return value
+    if math.isnan(value) or math.isinf(value) or value < 0 or value > 1:
+        raise typer.BadParameter("must be within 0-1")
+    return value
+
+
 def _version_cb(value: bool):
     if value:
         console.print(f"ctxlens {__version__}")
@@ -56,11 +77,11 @@ def analyze(
     fmt: str = typer.Option("auto", "--format", "-f", help="Force a parser (auto detects)."),
     tokenizer: str = typer.Option("auto", "--tokenizer", "-t", help="Tokenizer: auto|heuristic|tiktoken."),
     as_json: bool = typer.Option(False, "--json", help="Emit JSON instead of a terminal report."),
-    top: int = typer.Option(10, "--top", help="Number of top consumers to compute."),
-    tool_result_cap: int = typer.Option(400, "--tool-result-cap", help="Per tool-result token cap."),
-    tool_def_budget: int = typer.Option(800, "--tool-def-budget", help="Tool-definitions token budget."),
+    top: int = typer.Option(10, "--top", help="Number of top consumers to compute.", callback=_positive_int),
+    tool_result_cap: int = typer.Option(400, "--tool-result-cap", help="Per tool-result token cap.", callback=_positive_int),
+    tool_def_budget: int = typer.Option(800, "--tool-def-budget", help="Tool-definitions token budget.", callback=_non_negative_int),
     fail_over: float | None = typer.Option(
-        None, "--fail-over-ratio", help="Exit non-zero if waste ratio exceeds this (0-1). CI-friendly."
+        None, "--fail-over-ratio", help="Exit non-zero if waste ratio exceeds this (0-1). CI-friendly.", callback=_ratio
     ),
 ):
     """Analyze a single transcript and print a context profile."""
@@ -81,9 +102,9 @@ def report(
     out: Path | None = typer.Option(None, "--out", "-o", help="Write report to this file."),
     fmt: str = typer.Option("auto", "--format", "-f"),
     tokenizer: str = typer.Option("auto", "--tokenizer", "-t"),
-    tool_result_cap: int = typer.Option(400, "--tool-result-cap"),
-    tool_def_budget: int = typer.Option(800, "--tool-def-budget"),
-    fail_over: float | None = typer.Option(None, "--fail-over-ratio"),
+    tool_result_cap: int = typer.Option(400, "--tool-result-cap", callback=_positive_int),
+    tool_def_budget: int = typer.Option(800, "--tool-def-budget", callback=_non_negative_int),
+    fail_over: float | None = typer.Option(None, "--fail-over-ratio", callback=_ratio),
 ):
     """Generate an HTML (or JSON) report, to a file or stdout."""
     analysis = _load(path, fmt, tokenizer, 10, tool_result_cap, tool_def_budget)
diff --git a/tests/test_cli.py b/tests/test_cli.py
index 19cc563..4611011 100644
--- a/tests/test_cli.py
+++ b/tests/test_cli.py
@@ -92,6 +92,25 @@ def test_analyze_missing_file():
     assert result.exit_code == 1
 
 
+@pytest.mark.parametrize(
+    ("args", "reason"),
+    [
+        (["--top", "-1"], "must be a positive integer (>= 1)"),
+        (["--top", "0"], "must be a positive integer (>= 1)"),
+        (["--tool-result-cap", "-5"], "must be a positive integer (>= 1)"),
+        (["--tool-def-budget", "-1"], "must be a non-negative integer (>= 0)"),
+        (["--fail-over-ratio", "5"], "must be within 0-1"),
+        (["--fail-over-ratio", "-0.5"], "must be within 0-1"),
+        (["--fail-over-ratio", "nan"], "must be within 0-1"),
+    ],
+)
+def test_analyze_rejects_out_of_range_numeric_options(claude_jsonl, args, reason):
+    result = runner.invoke(app, ["analyze", str(claude_jsonl), *args])
+    assert result.exit_code == 2
+    assert "Invalid value" in result.stderr
+    assert reason in result.stderr
+
+
 def test_analyze_stdin(openai_array):
     raw = openai_array.read_text()
     result = runner.invoke(app, ["analyze", "-", "--json"], input=raw)