diff --git a/features/issue_6323_coverage_boost.feature b/features/issue_6323_coverage_boost.feature
new file mode 100644
index 000000000..b228c6128
--- /dev/null
+++ b/features/issue_6323_coverage_boost.feature
@@ -0,0 +1,56 @@
+Feature: Issue 6323 coverage — _format_size edge cases and formatting helper paths
+
+  Background:
+    Given a project context CLI in-memory database is initialized
+
+  @tdd_issue @tdd_issue_6323
+  Scenario Outline: _format_size covers all size display branches
+    When I invoke the _format_size helper with "<input>"
+    Then the _format_size result should be "<expected>"
+
+    Examples:
+      | input         | expected    |
+      | None          | (no limit)  |
+      | 0             | 0 bytes     |
+      | 1             | 1 byte      |
+      | 500           | 500 bytes   |
+      | 1024          | 1 KB        |
+      | 1536          | 1.5 KB      |
+      | 1572864       | 1.5 MB      |
+      | 1073741824    | 1 GB        |
+      | 2684354560    | 2.5 GB      |
+      | 1099511627776 | 1 TB        |
+      | 1649267441664 | 1.5 TB      |
+
+  @tdd_issue @tdd_issue_6323
+  Scenario: format_output wraps a dict message correctly in the json envelope
+    When I invoke format_output json with dict message level "warn" text "dict-msg-test"
+    Then the json envelope contains dict message with level "warn" text "dict-msg-test"
+
+  @tdd_issue @tdd_issue_6323
+  Scenario: format_output raises ValueError for a dict message missing required keys
+    When I invoke format_output json with malformed dict message
+    Then a ValueError is raised for the malformed message
+
+  @tdd_issue @tdd_issue_6323
+  Scenario: format_output plain output with empty-level dict message omits bracket prefix
+    When I invoke format_output plain with empty-level dict message text "no-prefix-plain"
+    Then the plain output contains "no-prefix-plain" without bracket prefix
+
+  @tdd_issue @tdd_issue_6323
+  Scenario: context show rich renders Execution Environment panel when policy blob has env set
+    Given a project "local/cov-ee" exists for context CLI
+    And issue 6323 project context sample is configured for "local/cov-ee"
+    And the context tier service has sample usage data for "local/cov-ee"
+    When I run context show rich on "local/cov-ee" with execution environment and integer depth gradient
+    Then the context show command should succeed
+    And the context show output contains "Execution Environment"
+
+  @tdd_issue @tdd_issue_6323
+  Scenario: context show rich handles depth gradient keys that cannot be parsed as integers
+    Given a project "local/cov-dg" exists for context CLI
+    And issue 6323 project context sample is configured for "local/cov-dg"
+    And the context tier service has sample usage data for "local/cov-dg"
+    When I run context show rich on "local/cov-dg" with non-integer depth gradient keys
+    Then the context show command should succeed
+    And the context show output contains "ACMS Pipeline"
diff --git a/features/steps/project_context_cli_steps.py b/features/steps/project_context_cli_steps.py
index 8362e15b7..982cae682 100644
--- a/features/steps/project_context_cli_steps.py
+++ b/features/steps/project_context_cli_steps.py
@@ -797,3 +797,201 @@ def step_issue6323_assert_json(context: Any, project: str, view: str) -> None:
 def step_issue6323_assert_yaml(context: Any, project: str, view: str) -> None:
     parsed = _issue6323_parse_yaml(context)
     _issue6323_assert_payload(parsed, context.issue6323_expected, project, view)
+
+
+# --------------------------------------------------------------------------
+# Issue 6323 coverage boost — _format_size edge cases
+# --------------------------------------------------------------------------
+
+
+@when('I invoke the _format_size helper with "{input_val}"')
+def step_invoke_format_size_helper(context: Any, input_val: str) -> None:
+    from cleveragents.cli.commands.project_context import _format_size
+
+    val: int | None = None if input_val == "None" else int(input_val)
+    context.format_size_result = _format_size(val)
+
+
+@then('the _format_size result should be "{expected}"')
+def step_assert_format_size_helper_result(context: Any, expected: str) -> None:
+    assert context.format_size_result == expected, (
+        f"Expected {expected!r}, got {context.format_size_result!r}"
+    )
+
+
+# --------------------------------------------------------------------------
+# Issue 6323 coverage boost — format_output dict message paths
+# --------------------------------------------------------------------------
+
+
+@when('I invoke format_output json with dict message level "{level}" text "{text}"')
+def step_invoke_format_output_dict_msg(context: Any, level: str, text: str) -> None:
+    import io
+    import sys
+
+    from cleveragents.cli.formatting import format_output
+
+    buf = io.StringIO()
+    old = sys.stdout
+    sys.stdout = buf
+    try:
+        format_output(
+            {"k": "v"},
+            "json",
+            command="cov-test",
+            status="ok",
+            exit_code=0,
+            messages=[{"level": level, "text": text}],
+        )
+    finally:
+        sys.stdout = old
+    context.cov_json_output = buf.getvalue()
+
+
+@then('the json envelope contains dict message with level "{level}" text "{text}"')
+def step_assert_json_envelope_dict_msg(context: Any, level: str, text: str) -> None:
+    parsed = json.loads(context.cov_json_output.strip())
+    msgs = parsed.get("messages", [])
+    assert any(
+        isinstance(m, dict) and m.get("level") == level and m.get("text") == text
+        for m in msgs
+    ), f"Expected dict message with level={level!r} text={text!r} in {msgs!r}"
+
+
+@when("I invoke format_output json with malformed dict message")
+def step_invoke_format_output_malformed_dict(context: Any) -> None:
+    import io
+    import sys
+
+    from cleveragents.cli.formatting import format_output
+
+    buf = io.StringIO()
+    old = sys.stdout
+    sys.stdout = buf
+    context.cov_raised = None
+    try:
+        format_output(
+            {"k": "v"},
+            "json",
+            command="cov-test",
+            status="ok",
+            exit_code=0,
+            messages=[{"bad_key": "missing level and text"}],
+        )
+    except ValueError as exc:
+        context.cov_raised = str(exc)
+    finally:
+        sys.stdout = old
+
+
+@then("a ValueError is raised for the malformed message")
+def step_assert_value_error_malformed_msg(context: Any) -> None:
+    assert context.cov_raised is not None, "Expected ValueError but none was raised"
+
+
+@when('I invoke format_output plain with empty-level dict message text "{text}"')
+def step_invoke_format_output_plain_empty_level(context: Any, text: str) -> None:
+    import io
+    import sys
+
+    from cleveragents.cli.formatting import format_output
+
+    buf = io.StringIO()
+    old = sys.stdout
+    sys.stdout = buf
+    try:
+        format_output(
+            {},
+            "plain",
+            command="cov-test",
+            status="ok",
+            exit_code=0,
+            messages=[{"level": "", "text": text}],
+        )
+    finally:
+        sys.stdout = old
+    context.cov_plain_output = buf.getvalue()
+
+
+@then('the plain output contains "{text}" without bracket prefix')
+def step_assert_plain_no_bracket_prefix(context: Any, text: str) -> None:
+    output = context.cov_plain_output
+    assert text in output, f"Expected {text!r} in plain output {output!r}"
+    for line in output.strip().split("\n"):
+        if text in line:
+            assert not line.startswith("["), (
+                f"Line should not start with '[': {line!r}"
+            )
+
+
+# --------------------------------------------------------------------------
+# Issue 6323 coverage boost — rich output paths (execution env, depth gradient)
+# --------------------------------------------------------------------------
+
+_COV_ACMS_BASE: dict[str, Any] = {
+    "hot_max_tokens": 8192,
+    "warm_max_decisions": 50,
+    "cold_max_decisions": 200,
+    "summarize": True,
+    "temporal_scope": "current",
+    "auto_refresh": True,
+    "default_breadth": 3,
+    "default_depth": 3,
+    "skeleton_ratio": 0.15,
+    "strategies": [],
+}
+
+
+@when(
+    'I run context show rich on "{project}" with execution environment'
+    " and integer depth gradient"
+)
+def step_ctx_show_rich_ee_int_gradient(context: Any, project: str) -> None:
+    from cleveragents.cli.commands.project_context import context_show
+
+    acms = dict(_COV_ACMS_BASE)
+    acms["depth_gradient"] = {"1": 0.8, "2": 0.5}
+    with (
+        patch(
+            "cleveragents.cli.commands.project_context._load_policy_json",
+            return_value={
+                "execution_environment": "host",
+                "execution_env_priority": "high",
+            },
+        ),
+        patch(
+            "cleveragents.cli.commands.project_context._read_acms_config",
+            return_value=acms,
+        ),
+    ):
+        _run_with_container(
+            context, context_show, project=project, view=None, output_format="rich"
+        )
+
+
+@when('I run context show rich on "{project}" with non-integer depth gradient keys')
+def step_ctx_show_rich_str_gradient(context: Any, project: str) -> None:
+    from cleveragents.cli.commands.project_context import context_show
+
+    acms = dict(_COV_ACMS_BASE)
+    acms["depth_gradient"] = {"hop_a": 0.8, "hop_b": 0.5}
+    with (
+        patch(
+            "cleveragents.cli.commands.project_context._load_policy_json",
+            return_value=None,
+        ),
+        patch(
+            "cleveragents.cli.commands.project_context._read_acms_config",
+            return_value=acms,
+        ),
+    ):
+        _run_with_container(
+            context, context_show, project=project, view=None, output_format="rich"
+        )
+
+
+@then('the context show output contains "{text}"')
+def step_ctx_show_output_contains(context: Any, text: str) -> None:
+    assert text in context.ctx_output, (
+        f"Expected {text!r} in context show output.\nGot:\n{context.ctx_output}"
+    )