diff --git a/src/agent-sec-core/agent-sec-cli/src/agent_sec_cli/skill_ledger/cli.py b/src/agent-sec-core/agent-sec-cli/src/agent_sec_cli/skill_ledger/cli.py
index e12f33a1b6..06f15b103d 100644
--- a/src/agent-sec-core/agent-sec-cli/src/agent_sec_cli/skill_ledger/cli.py
+++ b/src/agent-sec-core/agent-sec-cli/src/agent_sec_cli/skill_ledger/cli.py
@@ -33,7 +33,10 @@
         "  drifted   Skill files changed since last certification\n"
         "  warn      Scan found low-risk issues\n"
         "  deny      Scan found high-risk issues\n"
-        "  tampered  Ledger metadata is incomplete or fails authenticity checks"
+        "  tampered  Ledger metadata is incomplete or fails authenticity checks\n"
+        "  unmanaged Skill directory is not covered by any managed skill\n"
+        "            directory, or ledger state is not writable\n"
+        "  error     Check could not be completed for the skill"
     ),
     add_completion=True,
 )
@@ -186,6 +189,10 @@ def cmd_check(
       warn      Signature valid, but scan found low-risk issues
       deny      Signature valid, but scan found high-risk issues
       tampered  Ledger metadata is incomplete or fails authenticity checks
+      error     Check could not run for the skill (e.g. missing SKILL.md);
+                single checks print a JSON object with status and error
+                keys and exit 1, while --all records it as a per-skill
+                entry and checks the rest
 
     Use --all to check every registered skill and receive a JSON array of
     enriched results. Skill discovery uses built-in default directories plus
diff --git a/src/agent-sec-core/tests/unit-test/skill_ledger/test_cli_help.py b/src/agent-sec-core/tests/unit-test/skill_ledger/test_cli_help.py
new file mode 100644
index 0000000000..f10444d21e
--- /dev/null
+++ b/src/agent-sec-core/tests/unit-test/skill_ledger/test_cli_help.py
@@ -0,0 +1,64 @@
+"""Help-text contract tests for the ``skill-ledger`` command group."""
+
+import json
+import re
+from pathlib import Path
+
+from agent_sec_cli.security_middleware.backends.skill_ledger import (
+    _VERDICT_SEVERITY,
+)
+from agent_sec_cli.skill_ledger.cli import app
+from typer.testing import CliRunner
+
+# rich emits ANSI color codes when the environment forces color on (e.g.
+# GITHUB_ACTIONS or FORCE_COLOR), so strip them before matching on layout.
+_ANSI_ESCAPE_RE = re.compile(r"\x1b\[[0-9;]*m")
+
+
+def _plain_lines(output: str) -> list[str]:
+    """Return help output as ANSI-free lines with indentation stripped."""
+    return [_ANSI_ESCAPE_RE.sub("", line).strip() for line in output.splitlines()]
+
+
+def _has_status_line(output: str, status: str) -> bool:
+    """True if any line starts with ``status`` followed by a word boundary.
+
+    Rich renders help differently per environment (ANSI codes, 2- vs
+    3-space indent, width padding), so the match is layout-independent.
+    """
+    return any(re.match(rf"^{status}\b", line) for line in _plain_lines(output))
+
+
+def test_app_help_lists_every_verdict_status() -> None:
+    result = CliRunner().invoke(app, ["--help"])
+
+    assert result.exit_code == 0
+    for status in _VERDICT_SEVERITY:
+        assert _has_status_line(
+            result.output, status
+        ), f"status {status!r} missing from skill-ledger --help"
+
+
+def test_check_help_lists_error_status() -> None:
+    result = CliRunner().invoke(app, ["check", "--help"])
+
+    assert result.exit_code == 0
+    assert _has_status_line(
+        result.output, "error"
+    ), "check --help must document the error status"
+    # error is produced by both single-check failures and batch entries;
+    # the old "(--all path only)" qualifier was inaccurate.
+    assert "(--all path only)" not in result.output
+    # check never returns unmanaged (only show does, via manageability checks).
+    assert not _has_status_line(result.output, "unmanaged")
+
+
+def test_check_missing_dir_prints_error_status_json(tmp_path: Path) -> None:
+    result = CliRunner().invoke(app, ["check", str(tmp_path / "missing")])
+
+    assert result.exit_code == 1
+    payload = json.loads(result.output)
+    # Assert only the error envelope contract; the message text is free to
+    # evolve as new failure types are added.
+    assert payload["status"] == "error"
+    assert payload["error"]