Added Kimi K2 provider tests Fixes #40 #192
Conversation
|
Warning Rate limit exceeded@Sahilbhatane has exceeded the limit for the number of commits or files that can be reviewed per hour. Please wait 11 minutes and 29 seconds before requesting another review. ⌛ How to resolve this issue?After the wait time has elapsed, a review can be triggered using the We recommend that you space out your commits to avoid hitting the rate limit. 🚦 How do rate limits work?CodeRabbit enforces hourly rate limits for each developer per organization. Our paid plans have higher rate limits than the trial, open-source and free plans. In all cases, we re-allow further reviews after a brief timeout. Please see our FAQ for further information. 📒 Files selected for processing (3)
Note Other AI code review bot(s) detectedCodeRabbit has detected other AI code review bot(s) in this pull request and will avoid duplicating their findings in the review comments. This may lead to a less comprehensive review. WalkthroughAdds multi-provider NL→shell translation (OpenAI, Anthropic, Kimi K2, FAKE), provider-aware CLI API-key selection and fake-mode handling, a reworked preferences manager with serialization, Docker-backed integration test utilities and end-to-end tests, and updates to packaging, requirements, and docs. Changes
Sequence Diagram(s)sequenceDiagram
participant CLI as Cortex CLI
participant Interpreter as CommandInterpreter
participant SDK as OpenAI/Anthropic SDKs
participant KimiAPI as Kimi HTTP API
CLI->>Interpreter: NL→shell request (provider selected)
alt provider == kimi
Interpreter->>KimiAPI: POST structured payload (system+user+model, headers)
KimiAPI-->>Interpreter: assistant response / error
Interpreter->>Interpreter: parse_with_context → _parse_commands → _validate_commands
Interpreter-->>CLI: commands or error
else provider == openai/anthropic
Interpreter->>SDK: SDK call (model, system prompt)
SDK-->>Interpreter: assistant response
Interpreter->>Interpreter: parse_with_context → _parse_commands → _validate_commands
Interpreter-->>CLI: commands or error
else provider == fake
Interpreter->>Interpreter: load `CORTEX_FAKE_COMMANDS` or env/default
Interpreter-->>CLI: deterministic commands
end
Estimated code review effort🎯 4 (Complex) | ⏱️ ~50–70 minutes
Possibly related PRs
Suggested labels
Suggested reviewers
Poem
Pre-merge checks and finishing touches❌ Failed checks (3 warnings, 1 inconclusive)
✅ Passed checks (1 passed)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 3
🧹 Nitpick comments (20)
cortex/__init__.py (1)
1-2: Avoid duplicating the version string in multiple places.
__version__ = "0.1.0"here andversion="0.1.0"insetup.pywill drift over time. Consider defining the version in a single module (e.g.,cortex/_version.py) and importing it both here and insetup.py.setup.py (1)
7-8: Make requirements parsing robust to indented comments.Current list comprehension:
requirements = [line.strip() for line in fh if line.strip() and not line.startswith("#")]will include lines like
" # comment"as a requirement becauseline.startswith("#")ignores leading whitespace. Consider:stripped = line.strip() if stripped and not stripped.startswith("#"): requirements.append(stripped)to avoid accidentally treating indented comments as dependencies.
cortex/coordinator.py (5)
12-39: Use explicitNonechecks indurationfor clarity.To avoid relying on truthiness of floats and to be explicit about the sentinel value, consider:
def duration(self) -> Optional[float]: if self.start_time is not None and self.end_time is not None: return self.end_time - self.start_time return NoneThis is a minor readability/robustness improvement.
87-97: Avoid fully silent failures in_log.Right now any error opening/writing the log file is swallowed:
try: ... except Exception: passConsider at least emitting a minimal
print(..., file=sys.stderr)or narrowing the exception type so that unexpected errors don’t fail silently. This keeps installation resilient while still giving a hint when logging is misconfigured.
99-141: Be explicit aboutshell=Trueand broad exception handling trade-offs.Given these commands are arbitrary shell lines,
shell=Trueis probably required, but:
- It’s a security footgun if ever exposed to untrusted input or run under elevated privileges.
- Catching bare
Exceptionhere may hide unexpected programmer errors.At minimum, consider documenting that
InstallationCoordinatoris intended for local, trusted environments and not for executing untrusted commands. If feasible, narrow theexcept Exceptionto specific failures (e.g.,OSError) so unexpected errors surface during development.
166-197: Progress callback is invoked before status is updated, so callers never see final state.
progress_callbackis called before_execute_command, whilestep.statusis stillPENDING. The CLI’s callback usesstep.statusto print[OK]/[FAIL], so it will only ever see[PENDING].Consider instead:
for i, step in enumerate(self.steps): if self.progress_callback: self.progress_callback(i + 1, len(self.steps), step) # optional “starting” hook success = self._execute_command(step) if self.progress_callback: self.progress_callback(i + 1, len(self.steps), step) # “completed” hook with final statusor move the callback to after
_execute_commandif you only want the completed status. This will make CLI progress output reflect real step outcomes.
269-307:install_dockeris tightly coupled to Debian/Ubuntu + systemd environments.The hard-coded
aptcommands,add-apt-repository, andsystemctlchecks make this helper effectively Ubuntu/Debian with systemd only. That’s fine for container-based tests, but if you plan to reuse it on other distros (e.g., Arch, Fedora, non-systemd setups), it may fail in surprising ways.Consider either documenting this assumption clearly or wiring distro detection into the higher-level CLI so alternate installers can be plugged in later.
test/run_all_tests.py (1)
11-33: Test discovery correctly scoped totest/but won’t pick tests outside
discover_tests()starts fromos.path.dirname(__file__), so only tests under thetestdirectory are discovered, which matches howcortex/cli.pylocatesrun_all_tests.py. If you ever wantLLM/test_interpreter.py(or other top-level tests) to be run via this entry point, you’ll need to broadenstart_dir(e.g., repo root) or mirror those tests undertest/.README.md (1)
93-95: Contribution and testing references are helpful—verify link targetsPointing contributors to Issues,
contribution.md, andtest.mdis good for onboarding and testing discipline. Just double-check that these relative links resolve correctly in the GitHub UI from this README.LLM/test_interpreter.py (1)
97-103: Remove redundant@patch('openai.OpenAI')to satisfy Ruff and simplify the testIn
test_parse_empty_input, themock_openaiargument is unused, andsetUpalready stubs theopenaimodule globally. You can safely drop the patch decorator and argument—this will also resolve Ruff’sARG002warning.- @patch('openai.OpenAI') - def test_parse_empty_input(self, mock_openai): + def test_parse_empty_input(self): interpreter = CommandInterpreter(api_key=self.api_key, provider="openai") with self.assertRaises(ValueError): interpreter.parse("")test/test_cli.py (2)
32-37: Strengthentest_get_api_key_not_foundby asserting error outputYou already patch
sys.stderrhere, but the test only checks that the API key isNone, andmock_stderris otherwise unused (Ruff ARG002).To fully exercise
_get_api_key’s error path and use the mock, consider also asserting that something was written to stderr, e.g.:@patch.dict(os.environ, {}, clear=True) @patch('sys.stderr') def test_get_api_key_not_found(self, mock_stderr) -> None: - api_key = self.cli._get_api_key('openai') - self.assertIsNone(api_key) + api_key = self.cli._get_api_key('openai') + self.assertIsNone(api_key) + self.assertTrue(mock_stderr.write.called)This both validates the user-visible error and clears the static-analysis warning.
90-196: Avoid real spinner sleeps in install-path unit tests (optional)All install tests that reach
CortexCLI.installwith a valid API key will run the 10‑iteration spinner loop, each iteration sleeping for 0.1s. As the number of tests grows, this can noticeably slow the suite.If runtime becomes an issue, consider patching
_animate_spinnerfor these tests:@patch.object(CortexCLI, "_animate_spinner", return_value=None) @patch.dict(os.environ, {'OPENAI_API_KEY': 'test-key'}, clear=True) @patch('cortex.cli.CommandInterpreter') def test_install_dry_run(self, mock_interpreter_class, _mock_spinner) -> None: ...This keeps behaviour identical from the CLI’s perspective while making tests faster and less timing‑sensitive.
cortex/test_coordinator.py (1)
36-321: Coordinator tests look comprehensive; consider makingtest_step_timingmore robustThe coordinator test suite here is thorough and matches the coordinator’s behaviour well (init, stop/continue on error, rollback, logging, verification, summaries, and JSON export).
One minor robustness point: in
test_step_timing, you assertstep.end_time > step.start_time. On platforms with coarsetime.time()resolution, two consecutive calls can occasionally return the same value, which would make this test intermittently fail even though the coordinator is behaving correctly.If you want to eliminate that flakiness risk, you could either:
- Relax to
>=or- Patch
time.time()to return controlled, strictly increasing values during the test.Otherwise, this module looks solid.
LLM/interpreter.py (1)
154-185: Minor cleanup in_call_fake: avoid double JSON parsing ofCORTEX_FAKE_COMMANDSThe fake provider path correctly validates that
CORTEX_FAKE_COMMANDSis JSON with a top‑level"commands"key, but then discards the parseddataand calls_parse_commands(payload), which re‑parses the same JSON string.Functionally this is fine, but you can simplify and avoid double parsing by reusing
data, for example:payload = os.environ.get("CORTEX_FAKE_COMMANDS") if payload: try: - data = json.loads(payload) + data = json.loads(payload) except json.JSONDecodeError as exc: raise ValueError("CORTEX_FAKE_COMMANDS must contain valid JSON") from exc - if not isinstance(data, dict) or "commands" not in data: - raise ValueError("CORTEX_FAKE_COMMANDS must define a 'commands' list") - return self._parse_commands(payload) + if not isinstance(data, dict) or "commands" not in data: + raise ValueError("CORTEX_FAKE_COMMANDS must define a 'commands' list") + commands = data["commands"] + if not isinstance(commands, list): + raise ValueError("CORTEX_FAKE_COMMANDS 'commands' must be a list") + return [cmd for cmd in commands if cmd and isinstance(cmd, str)]
parse()will still run_validate_commandson this list, so you keep the safety guarantees while reducing redundant work.test/integration/docker_utils.py (1)
55-108: Mounted paths are documented as read‑only but are actually read‑writeIn
run_in_docker()themountsparameter is documented as:“Iterable of host
Pathinstances mounted read-only to the same location within the container.”However, the volume flag is built as:
docker_cmd.extend([ "-v", f"{str(host_path.resolve())}:{container_path}", ])Without a
:rosuffix, these mounts are read‑write by default. For integration tests this may be intentional (e.g., writing artifacts back to the host), but it contradicts the docstring and slightly weakens isolation.Consider either:
- Appending
:roto the volume spec to enforce read‑only mounts, or- Updating the docstring to reflect that mounts are read‑write and, if needed, adding an option to request read‑only behaviour.
cortex/cli.py (5)
22-36: Provider fallback could cause confusing error messages.Line 36 returns
'openai'as the default even when noOPENAI_API_KEYis present. This will cause a delayed error in_get_api_keyrather than failing fast with a clear message that no provider is configured.Consider one of these approaches:
Option 1: Return None and handle in caller
return 'fake' - return 'openai' + return None # No provider configuredThen check in the
installmethod:provider = self._get_provider() if provider is None: self._print_error("No LLM provider configured. Set OPENAI_API_KEY, ANTHROPIC_API_KEY, or KIMI_API_KEY.") return 1Option 2: Emit clear error immediately
return 'fake' - return 'openai' + self._print_error("No LLM provider configured. Set one of: OPENAI_API_KEY, ANTHROPIC_API_KEY, or KIMI_API_KEY.") + sys.exit(1)
92-108: Consider removing redundant "install" prefix.Line 103 prefixes the software parameter with
"install "when calling the interpreter. Since the CLI command is alreadyinstalland the parameter is namedsoftware, this might be redundant or confusing. Users might naturally include verbs in their input (e.g.,cortex install "set up docker").Consider passing the software parameter directly:
- commands = interpreter.parse(f"install {software}") + commands = interpreter.parse(software)Or make the prompt more explicit:
- commands = interpreter.parse(f"install {software}") + commands = interpreter.parse(f"Generate shell commands to install {software}")
118-156: Execute mode logic is sound, but consider adding else block.The execution flow with progress callbacks and error handling is well-structured. However, Lines 153-155 should be in an
elseblock for clarity, as flagged by static analysis (TRY300).Apply this diff to improve structure:
print(f" Error: {result.error_message}", file=sys.stderr) return 1 + else: + print("\nTo execute these commands, run with --execute flag") + print("Example: cortex install docker --execute") + return 0 - print("\nTo execute these commands, run with --execute flag") - print("Example: cortex install docker --execute") - return 0 - except ValueError as exc:
157-166: Exception handling is appropriate for CLI entry point.The tiered exception handling (ValueError for input errors, RuntimeError for API failures, and Exception as fallback) is suitable for a CLI entry point where graceful error messages are important. The blind
Exceptioncatch on Line 163 prevents unexpected tracebacks from reaching end users, which is acceptable in this context.Static analysis suggests using explicit conversion flags in f-strings (RUF010):
- self._print_error(f"API call failed: {str(exc)}") + self._print_error(f"API call failed: {exc!s}")- self._print_error(f"Unexpected error: {str(exc)}") + self._print_error(f"Unexpected error: {exc!s}")These are minor style improvements and not critical.
197-198: Consider making --execute and --dry-run mutually exclusive.Both flags can currently be specified together, and
dry_runtakes precedence (checked first on Line 114). This could confuse users who expect both flags to trigger an error.Make the flags mutually exclusive:
install_parser = subparsers.add_parser('install', help='Install software using natural language') install_parser.add_argument('software', type=str, help='Software to install (natural language)') - install_parser.add_argument('--execute', action='store_true', help='Execute the generated commands') - install_parser.add_argument('--dry-run', action='store_true', help='Show commands without executing') + + exec_group = install_parser.add_mutually_exclusive_group() + exec_group.add_argument('--execute', action='store_true', help='Execute the generated commands') + exec_group.add_argument('--dry-run', action='store_true', help='Show commands without executing')
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (19)
.gitignore(1 hunks)LLM/interpreter.py(9 hunks)LLM/requirements.txt(1 hunks)LLM/test_interpreter.py(11 hunks)MANIFEST.in(1 hunks)README.md(2 hunks)contribution.md(1 hunks)cortex/__init__.py(1 hunks)cortex/cli.py(1 hunks)cortex/coordinator.py(1 hunks)cortex/test_cli.py(1 hunks)cortex/test_coordinator.py(1 hunks)setup.py(1 hunks)test.md(1 hunks)test/integration/__init__.py(1 hunks)test/integration/docker_utils.py(1 hunks)test/integration/test_end_to_end.py(1 hunks)test/run_all_tests.py(1 hunks)test/test_cli.py(1 hunks)
🧰 Additional context used
🧬 Code graph analysis (9)
test/run_all_tests.py (1)
cortex/cli.py (1)
main(168-222)
cortex/cli.py (3)
LLM/interpreter.py (2)
CommandInterpreter(18-268)parse(224-243)cortex/coordinator.py (3)
InstallationCoordinator(53-266)StepStatus(12-19)execute(166-213)test/run_all_tests.py (1)
main(19-33)
LLM/test_interpreter.py (1)
LLM/interpreter.py (4)
CommandInterpreter(18-268)APIProvider(9-15)parse(224-243)_call_kimi(118-152)
test/integration/test_end_to_end.py (3)
test/integration/docker_utils.py (3)
docker_available(26-52)run_in_docker(55-108)succeeded(21-23)cortex/cli.py (1)
main(168-222)test/run_all_tests.py (1)
main(19-33)
cortex/test_coordinator.py (1)
cortex/coordinator.py (11)
InstallationCoordinator(53-266)InstallationStep(23-39)InstallationResult(43-50)StepStatus(12-19)install_docker(269-307)duration(35-39)execute(166-213)add_rollback_command(162-164)verify_installation(215-237)get_summary(239-261)export_log(263-266)
cortex/test_cli.py (1)
cortex/cli.py (9)
CortexCLI(14-165)main(168-222)_get_api_key(38-54)_get_provider(22-36)_print_status(56-58)_print_error(60-62)_print_success(64-66)install(80-165)_animate_spinner(68-73)
cortex/coordinator.py (1)
cortex/cli.py (1)
progress_callback(119-126)
cortex/__init__.py (1)
cortex/cli.py (1)
main(168-222)
test/test_cli.py (1)
cortex/cli.py (9)
CortexCLI(14-165)main(168-222)_get_api_key(38-54)_get_provider(22-36)_print_status(56-58)_print_error(60-62)_print_success(64-66)install(80-165)_animate_spinner(68-73)
🪛 Ruff (0.14.4)
test/integration/docker_utils.py
34-34: subprocess call: check for execution of untrusted input
(S603)
42-42: subprocess call: check for execution of untrusted input
(S603)
50-50: Consider moving this statement to an else block
(TRY300)
91-91: Use explicit conversion flag
Replace with conversion flag
(RUF010)
99-99: subprocess call: check for execution of untrusted input
(S603)
cortex/cli.py
155-155: Consider moving this statement to an else block
(TRY300)
161-161: Use explicit conversion flag
Replace with conversion flag
(RUF010)
163-163: Do not catch blind exception: Exception
(BLE001)
164-164: Use explicit conversion flag
Replace with conversion flag
(RUF010)
210-210: subprocess call: check for execution of untrusted input
(S603)
LLM/test_interpreter.py
98-98: Unused method argument: mock_openai
(ARG002)
cortex/test_cli.py
28-28: Unused method argument: mock_stderr
(ARG002)
48-48: Unused method argument: mock_stderr
(ARG002)
53-53: Unused method argument: mock_stdout
(ARG002)
cortex/coordinator.py
75-75: Avoid specifying long messages outside the exception class
(TRY003)
96-97: try-except-pass detected, consider logging the exception
(S110)
96-96: Do not catch blind exception: Exception
(BLE001)
107-107: subprocess call with shell=True identified, security issue
(S602)
136-136: Do not catch blind exception: Exception
(BLE001)
140-140: Use explicit conversion flag
Replace with conversion flag
(RUF010)
153-153: subprocess call with shell=True identified, security issue
(S602)
159-159: Do not catch blind exception: Exception
(BLE001)
160-160: Use explicit conversion flag
Replace with conversion flag
(RUF010)
223-223: subprocess call with shell=True identified, security issue
(S602)
233-233: Do not catch blind exception: Exception
(BLE001)
235-235: Use explicit conversion flag
Replace with conversion flag
(RUF010)
LLM/interpreter.py
55-55: Avoid specifying long messages outside the exception class
(TRY003)
146-146: Abstract raise to an inner function
(TRY301)
146-146: Avoid specifying long messages outside the exception class
(TRY003)
149-149: Abstract raise to an inner function
(TRY301)
149-149: Avoid specifying long messages outside the exception class
(TRY003)
152-152: Avoid specifying long messages outside the exception class
(TRY003)
152-152: Use explicit conversion flag
Replace with conversion flag
(RUF010)
162-162: Avoid specifying long messages outside the exception class
(TRY003)
164-164: Avoid specifying long messages outside the exception class
(TRY003)
227-227: Avoid specifying long messages outside the exception class
(TRY003)
test/test_cli.py
34-34: Unused method argument: mock_stderr
(ARG002)
🔇 Additional comments (28)
.gitignore (1)
1-30: Solid foundation for Python project hygiene.The
.gitignorefile appropriately covers Python bytecode, build artifacts, virtual environments, and testing/coverage outputs—all essential for the multi-provider LLM CLI and Docker-backed integration test infrastructure this PR introduces. Patterns are syntactically correct and follow standard conventions.MANIFEST.in (1)
1-5: Packaging manifest looks appropriate for the current project layout.README, LICENSE, Python sources, and LLM requirements are included; this is a reasonable minimal set for distribution.
contribution.md (1)
12-20: Confirmsrc/requirements.txtpath and keep install docs aligned.The setup instructions depend on
src/requirements.txt. Please double-check that this file exists in the repo and that other docs (e.g.,test.md, README) reference the same path so new contributors don’t hit a missing-file error.test.md (1)
1-67: Testing guide is clear and aligns with the described test suites.Unit vs integration responsibilities, Docker usage, and CI recommendations are well scoped and actionable; this should make it straightforward for contributors to run the right tests locally and in CI.
test/integration/__init__.py (1)
1-1: Clear package-level docstringThe docstring succinctly documents the purpose of the integration test package; no changes needed.
README.md (1)
61-76: Provider table and AI layer docs align with implementationThe updated AI stack line and the “Supported LLM Providers” table accurately reflect the current providers and defaults (
OPENAI_API_KEY,ANTHROPIC_API_KEY,KIMI_API_KEY,CORTEX_PROVIDER=fakewithCORTEX_FAKE_COMMANDS). This should make configuring OpenAI/Claude/Kimi/Fake much clearer for users.LLM/test_interpreter.py (7)
2-10: Test module setup is pragmatic and isolates interpreter dependenciesUsing
sys.path.insertplussys.modulesstubs foropenaiandanthropicis a practical way to exerciseCommandInterpreterwithout requiring real SDKs. This keeps tests self-contained and avoids network calls while still letting you assert initialization behavior.
15-24: Shared stubs insetUpkeep provider imports safe across testsThe
setUpmethod’spatch.dict(sys.modules, {"openai": ..., "anthropic": ...})ensures that every test uses stubs instead of real clients, which nicely centralizes mocking and avoids per-test boilerplate for import safety.
49-55: Kimi initialization test cleanly validates defaults and dependency handlingThe
test_initialization_kimicase, with@patch.dict(os.environ, {}, clear=True)and a stubbedrequestsmodule, correctly validates that the interpreter:
- Selects
APIProvider.KIMI- Falls back to the
kimi-k2default model whenKIMI_DEFAULT_MODELis unset- Requires
requeststo be importableThis is a solid pattern for providers that use plain HTTP clients rather than SDKs.
104-199: OpenAI/Claude call and parsing tests comprehensively cover core flowsThe tests around
_call_openai,_call_claude,parse, andparse_with_context(including success and failure branches, validation on/off, and context injection) exercise the main provider paths and the dangerous-command filtering well. They should catch most regressions in prompt formatting, JSON parsing, and validation logic.
221-242: Docker installation parsing test validates realistic multi-command output
test_parse_docker_installationensures that the interpreter returns a non-empty list and that at least one command mentions docker, using a realistic multi-command JSON payload. This is a good high-level sanity check on both parsing and basic intent preservation.
243-254: Fake provider tests nicely pin env-driven and default behaviorsThe FAKE provider tests:
- Confirm that
CORTEX_FAKE_COMMANDSJSON is honored exactly when set.- Fall back to the safe default docker command sequence when the env var is absent.
This gives you deterministic offline behavior and protects against regressions in the fake provider path.
255-284: Kimi HTTP tests correctly isolaterequestsand exercise success/error pathsThe Kimi tests stub
requestsviasys.modules, then:
- Validate happy-path JSON handling and command extraction in
_call_kimi.- Ensure any underlying exception from
postis surfaced as aRuntimeError.This pattern cleanly decouples tests from the real HTTP client and should make future refactors to Kimi integration safer.
test/integration/test_end_to_end.py (8)
1-21: LGTM! Well-structured test module setup.The module docstring, imports, and constants are well-organized. The Docker image selection via environment variable provides good flexibility, and the base environment variables are appropriate for containerized testing.
23-37: LGTM! Clean helper method for Docker execution.The test class properly skips when Docker is unavailable, and the
_runhelper method correctly merges environments and prepends pip bootstrapping. The use ofdict | Nonesyntax withfrom __future__ import annotationsensures compatibility.
39-44: LGTM! Effective smoke test.This test provides a good baseline check that the CLI entry point is functional in a clean container environment.
46-64: LGTM! Comprehensive dry-run test.The test properly configures the fake provider and validates that dry-run mode displays generated commands without executing them. The JSON structure correctly matches the expected format.
65-82: LGTM! Good coverage of execution flow.This test validates the execute path without requiring real API calls or executing potentially dangerous commands. The assertion checks for the expected success message.
83-99: LGTM! Good component-level test.The test properly isolates the coordinator component and uses secure heredoc syntax to prevent variable expansion. This validates the coordinator works correctly inside the container environment.
113-114: LGTM! Standard unittest main block.The main block follows standard Python testing patterns and is appropriately marked for coverage exclusion.
100-111: Assertion is valid and test is well-structured.The unittest runner correctly outputs "OK" in the last line of stdout, as verified. The assertion
self.assertIn("OK", result.stdout.splitlines()[-1])is sound and will reliably detect successful test runs even when additional details like skipped test counts are present.cortex/cli.py (7)
1-12: LGTM! Clean module setup.The imports and module docstring are well-organized and appropriate for the CLI implementation.
14-21: LGTM! Clean initialization.The spinner initialization is straightforward and uses appropriate Unicode braille patterns for visual feedback.
38-54: LGTM! Clean API key retrieval with helpful error messages.The method properly maps providers to environment variables and provides clear guidance when keys are missing.
56-79: LGTM! Well-designed output helpers.The helper methods provide clean abstractions for console output with appropriate use of stdout/stderr and ANSI escape sequences.
109-117: LGTM! Clear command display and dry-run handling.The generated commands are displayed in a user-friendly numbered format, and dry-run mode exits appropriately with helpful guidance.
202-211: LGTM! Test runner integration is secure.The test flag handling correctly constructs the path to the test runner and executes it using the current Python interpreter. The static analysis warning on Line 210 (S603) is a false positive—the
test_runnerpath is derived from__file__, not user input.
213-226: LGTM! Clean command dispatch and entry point.The command dispatch logic is straightforward, and the main block follows standard Python patterns.
|
@Sahilbhatane why are still solving multiple issues in single PR. Limit one issue for only one PR. |
|
hey @dhvll here is the video you asked - i will also share this on you discord. this video show CLI implantation and kimi k2 provider implementation as well. other details, reviews and conversation we will do on discord |
|
🎯 Closing for MVP Focus This issue is being closed to help the team focus on MVP-critical features (#1-45). This is NOT abandoned - it's an important feature we'll revisit after MVP completion. Timeline:
Want to work on this anyway? Tracking: Labeled as Thanks for understanding! 🚀 — Mike (@mikejmorgan-ai) |
Sahilbhatane
changed the title
There was a problem hiding this comment.
Actionable comments posted: 3
♻️ Duplicate comments (1)
LLM/interpreter.py (1)
262-271: Default model identifiers have been updated appropriately.The defaults have been updated from the previous review:
- OpenAI:
"gpt-4o"(updated from deprecated"gpt-4")- Kimi:
"kimi-k2-turbo-preview"with env override supportThis addresses the concerns from the previous review about deprecated model identifiers.
🧹 Nitpick comments (11)
test/test_coordinator.py (2)
5-5: Unused import.The
timemodule is imported but doesn't appear to be used in this file.-import time
319-319: PreferassertGreateroverassertTruewith comparison.As flagged by SonarCloud,
assertGreaterprovides better failure messages and is the idiomatic choice for this comparison.- self.assertTrue(step.end_time > step.start_time) + self.assertGreater(step.end_time, step.start_time)test/test_cli.py (1)
31-35: Consider prefixing unused mock argument.The
mock_stderrparameter is used only to suppress stderr output during the test. Prefixing with underscore signals intentional non-use.@patch.dict(os.environ, {}, clear=True) @patch('sys.stderr') - def test_get_api_key_not_found(self, mock_stderr): + def test_get_api_key_not_found(self, _mock_stderr): api_key = self.cli._get_api_key('openai') self.assertIsNone(api_key)setup.py (1)
6-6: Minor typo in comment.The comment has an extra
#character.-# # Try to read requirements from root, fallback to LLM directory +# Try to read requirements from root, fallback to LLM directorydocs/ISSUE_40_KIMI_K2_IMPLEMENTATION.md (1)
190-194: Add language specifier to fenced code block.The test output code block (lines 190-194) is missing a language specifier. Add
textorplaintextfor better markdown rendering.-``` +```text Ran 143 tests in 10.136s OK (skipped=5)</blockquote></details> <details> <summary>LLM/interpreter.py (1)</summary><blockquote> `135-136`: **Redundant import of `requests` inside `_call_kimi`.** The `requests` module is already imported and validated during `_initialize_client()` (line 53). This second import inside `_call_kimi` is unnecessary and the ImportError handling at line 152-153 will never trigger if initialization succeeded. ```diff try: - import requests response = requests.post( f"{self._kimi_base_url.rstrip('/')}/v1/chat/completions", headers=headers, json=payload, timeout=60, )Note: This also means removing the
ImportErrorcatch at lines 152-153 since initialization already guaranteesrequestsis available.cortex/user_preferences.py (5)
201-205: Silent exception swallowing hides initialization errors.The
try-except-passpattern silently ignores all exceptions during initial load. This can hide configuration corruption or permission issues.Consider logging the exception or at least catching a more specific exception type:
try: self.load() - except Exception: - # Ignore load errors here; callers can handle later - pass + except (IOError, ValueError) as e: + # Log but don't fail - callers can handle missing config + import logging + logging.debug(f"Initial config load skipped: {e}")
328-342: Unusedkey_partsparameter in_coerce_value_for_set.The
key_partsparameter is defined but never used in the method body. Either remove it or use it to provide context-aware coercion (e.g., knowing the target type based on the key path).- def _coerce_value_for_set(self, key_parts: List[str], value: Any) -> Any: + def _coerce_value_for_set(self, value: Any) -> Any: """Coerce string inputs into appropriate types for set()."""Then update the call site at line 361:
- coerced = self._coerce_value_for_set(parts, value) + coerced = self._coerce_value_for_set(value)
344-432:set()method has excessive cognitive complexity (56 vs. allowed 15).This method handles many different preference types with nested conditionals, making it difficult to maintain and test. Consider refactoring into a dispatch pattern or strategy-based approach.
One approach is to use a handler registry:
def set(self, key: str, value: Any) -> bool: if self._preferences is None: self.load() parts = key.split(".") coerced = self._coerce_value_for_set(value) handlers = { "verbosity": self._set_verbosity, "confirmations": self._set_confirmations, "auto_update": self._set_auto_update, "ai": self._set_ai, "packages": self._set_packages, "conflicts": self._set_conflicts, } handler = handlers.get(parts[0]) if handler: handler(parts, coerced, value) elif parts[0] in ["theme", "language", "timezone"]: setattr(self._preferences, parts[0], coerced) else: raise ValueError(f"Unknown preference key: {key}") self.save() return TrueThen extract each branch into a dedicated
_set_*method.
434-461:reset()always returnsTrue, even on success paths.The method signature promises a boolean return indicating success, but it always returns
Trueor raises an exception. This is fine semantically (exceptions for errors,Truefor success), but SonarCloud flags it because there's noFalsereturn path.If the intent is that errors always raise exceptions, consider changing the return type to
Noneor documenting that exceptions are the error signaling mechanism.
131-166:from_dicthas broad exception handling that masks errors.Lines 141-142 and 151-152 catch bare
Exceptionand silently default to fallback values. This can hide data corruption or invalid enum values without any indication to the caller.Consider catching the specific
ValueErrorthatEnum()raises:try: ai_data['creativity'] = AICreativity(ai_data['creativity']) - except Exception: + except ValueError: ai_data['creativity'] = AICreativity.BALANCEDSimilarly for verbosity at lines 149-152.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (11)
LLM/interpreter.py(9 hunks)LLM/requirements.txt(1 hunks)LLM/test_interpreter.py(5 hunks)cortex/cli.py(3 hunks)cortex/user_preferences.py(5 hunks)docs/ISSUE_40_KIMI_K2_IMPLEMENTATION.md(1 hunks)requirements-dev.txt(1 hunks)requirements.txt(1 hunks)setup.py(3 hunks)test/test_cli.py(3 hunks)test/test_coordinator.py(2 hunks)
✅ Files skipped from review due to trivial changes (1)
- requirements.txt
🚧 Files skipped from review as they are similar to previous changes (1)
- LLM/requirements.txt
🧰 Additional context used
🧬 Code graph analysis (4)
test/test_cli.py (1)
cortex/cli.py (3)
_get_api_key(47-63)_get_provider(31-45)install(84-212)
LLM/test_interpreter.py (1)
LLM/interpreter.py (3)
APIProvider(9-15)_call_kimi(118-155)parse(227-246)
test/test_coordinator.py (1)
cortex/coordinator.py (5)
InstallationCoordinator(65-350)InstallationStep(40-53)InstallationResult(57-62)StepStatus(31-36)install_docker(353-407)
cortex/cli.py (1)
cortex/user_preferences.py (1)
get(288-315)
🪛 GitHub Actions: Cortex Automation
cortex/user_preferences.py
[error] 12-12: ModuleNotFoundError: No module named 'yaml'. PyYAML is not installed; importing 'yaml' fails during test run.
🪛 GitHub Check: SonarCloud Code Analysis
test/test_coordinator.py
[warning] 319-319: Consider using "assertGreater" instead.
cortex/user_preferences.py
[warning] 328-328: Remove the unused function parameter "key_parts".
[failure] 344-344: Refactor this function to reduce its Cognitive Complexity from 56 to the 15 allowed.
[failure] 434-434: Refactor this method to not always return the same value.
🪛 LanguageTool
🪛 markdownlint-cli2 (0.18.1)
docs/ISSUE_40_KIMI_K2_IMPLEMENTATION.md
3-3: Bare URL used
(MD034, no-bare-urls)
4-4: Bare URL used
(MD034, no-bare-urls)
190-190: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
234-234: Bare URL used
(MD034, no-bare-urls)
235-235: Bare URL used
(MD034, no-bare-urls)
236-236: Bare URL used
(MD034, no-bare-urls)
🪛 Ruff (0.14.7)
test/test_cli.py
33-33: Unused method argument: mock_stderr
(ARG002)
80-80: Unused method argument: mock_history
(ARG002)
LLM/interpreter.py
55-55: Avoid specifying long messages outside the exception class
(TRY003)
147-147: Abstract raise to an inner function
(TRY301)
147-147: Avoid specifying long messages outside the exception class
(TRY003)
150-150: Abstract raise to an inner function
(TRY301)
150-150: Avoid specifying long messages outside the exception class
(TRY003)
153-153: Avoid specifying long messages outside the exception class
(TRY003)
155-155: Avoid specifying long messages outside the exception class
(TRY003)
155-155: Use explicit conversion flag
Replace with conversion flag
(RUF010)
165-165: Avoid specifying long messages outside the exception class
(TRY003)
167-167: Avoid specifying long messages outside the exception class
(TRY003)
230-230: Avoid specifying long messages outside the exception class
(TRY003)
cortex/user_preferences.py
141-141: Do not catch blind exception: Exception
(BLE001)
151-151: Do not catch blind exception: Exception
(BLE001)
203-205: try-except-pass detected, consider logging the exception
(S110)
203-203: Do not catch blind exception: Exception
(BLE001)
226-226: Consider moving this statement to an else block
(TRY300)
227-227: Do not catch blind exception: Exception
(BLE001)
228-228: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
228-228: Avoid specifying long messages outside the exception class
(TRY003)
228-228: Use explicit conversion flag
Replace with conversion flag
(RUF010)
250-250: Consider moving this statement to an else block
(TRY300)
253-253: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
253-253: Avoid specifying long messages outside the exception class
(TRY003)
253-253: Use explicit conversion flag
Replace with conversion flag
(RUF010)
254-254: Do not catch blind exception: Exception
(BLE001)
255-255: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
255-255: Avoid specifying long messages outside the exception class
(TRY003)
255-255: Use explicit conversion flag
Replace with conversion flag
(RUF010)
283-283: Consider moving this statement to an else block
(TRY300)
285-285: Do not catch blind exception: Exception
(BLE001)
286-286: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
286-286: Avoid specifying long messages outside the exception class
(TRY003)
286-286: Use explicit conversion flag
Replace with conversion flag
(RUF010)
328-328: Unused method argument: key_parts
(ARG002)
370-370: Do not catch blind exception: Exception
(BLE001)
371-371: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
371-371: Avoid specifying long messages outside the exception class
(TRY003)
375-375: Abstract raise to an inner function
(TRY301)
375-375: Avoid specifying long messages outside the exception class
(TRY003)
377-377: Abstract raise to an inner function
(TRY301)
377-377: Avoid specifying long messages outside the exception class
(TRY003)
382-382: Abstract raise to an inner function
(TRY301)
382-382: Avoid specifying long messages outside the exception class
(TRY003)
385-385: Abstract raise to an inner function
(TRY301)
385-385: Avoid specifying long messages outside the exception class
(TRY003)
387-387: Abstract raise to an inner function
(TRY301)
387-387: Avoid specifying long messages outside the exception class
(TRY003)
392-392: Abstract raise to an inner function
(TRY301)
392-392: Avoid specifying long messages outside the exception class
(TRY003)
399-399: Do not catch blind exception: Exception
(BLE001)
400-400: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
400-400: Avoid specifying long messages outside the exception class
(TRY003)
403-403: Abstract raise to an inner function
(TRY301)
403-403: Avoid specifying long messages outside the exception class
(TRY003)
410-410: Abstract raise to an inner function
(TRY301)
410-410: Avoid specifying long messages outside the exception class
(TRY003)
412-412: Abstract raise to an inner function
(TRY301)
412-412: Avoid specifying long messages outside the exception class
(TRY003)
417-417: Abstract raise to an inner function
(TRY301)
417-417: Avoid specifying long messages outside the exception class
(TRY003)
419-419: Abstract raise to an inner function
(TRY301)
419-419: Avoid specifying long messages outside the exception class
(TRY003)
426-426: Abstract raise to an inner function
(TRY301)
426-426: Avoid specifying long messages outside the exception class
(TRY003)
430-430: Consider moving this statement to an else block
(TRY300)
432-432: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
432-432: Avoid specifying long messages outside the exception class
(TRY003)
432-432: Use explicit conversion flag
Replace with conversion flag
(RUF010)
457-457: Avoid specifying long messages outside the exception class
(TRY003)
539-539: Avoid specifying long messages outside the exception class
(TRY003)
🔇 Additional comments (18)
LLM/test_interpreter.py (4)
15-24: LGTM!The setUp correctly patches
sys.modulesto stub external dependencies and ensures cleanup viaaddCleanup.
49-54: LGTM!The Kimi provider initialization test correctly verifies the default model
kimi-k2-turbo-previewis set when noKIMI_DEFAULT_MODELenvironment variable is present.
56-72: LGTM!The test properly verifies successful Kimi API calls, including the Authorization header format and command parsing.
74-84: LGTM!The failure test correctly verifies that HTTP errors are caught and re-raised as
RuntimeError.test/test_coordinator.py (1)
1-16: LGTM on import updates.The added imports for
Mock,patch,call,InstallationResult, andinstall_dockerare appropriate for the test file's needs.test/test_cli.py (3)
26-29: LGTM!Good addition of the KIMI provider API key test, maintaining consistency with the other provider tests.
47-55: LGTM!The KIMI provider detection and provider override tests correctly verify the
_get_provider()logic for KIMI and theCORTEX_PROVIDERoverride behavior.
77-88: LGTM!This test correctly verifies that the fake provider path bypasses API key validation and proceeds with command generation.
Note: The
mock_historyparameter (line 80) is unused but necessary to preventInstallationHistoryfrom performing file I/O. Consider prefixing with underscore (_mock_history) to signal intentional non-use.setup.py (3)
15-15: LGTM on fallback requirements.The fallback requirements list appropriately includes
requests>=2.32.4to support the new Kimi provider HTTP integration when no requirements file exists.
35-42: Verify Python 3.8 and 3.9 compatibility.The minimum Python version was lowered from 3.10 to 3.8. Ensure the codebase doesn't use Python 3.10+ features (e.g.,
matchstatements, union types with|syntax).
33-33: Verify license change is intentional.The license classifier was changed from
Apache Software LicensetoMIT License. Ensure this change is intentional and that the LICENSE file in the repository reflects this update.requirements-dev.txt (1)
10-10: Versionrequests>=2.32.4is valid.The
requests>=2.32.4dependency is a released PyPI package (includes security fix for CVE-2024-47081). However, verify that this dependency is not already included in the rootrequirements.txtreferenced at line 4 to avoid duplication.cortex/cli.py (4)
47-63: LGTM! API key retrieval is clean and well-structured.The provider-to-env-var mapping is correct and the error messaging guides users appropriately. Returning
Nonefor unknown providers (line 57) is acceptable since those would be handled elsewhere or caught in validation.
87-92: Fake provider handling is correct.The special case for
fakeprovider usingCORTEX_FAKE_API_KEYor the default'fake-api-key'is appropriate for testing scenarios. The early return on line 92 when the API key is missing prevents further execution with invalid credentials.
536-537: Documentation updates are helpful for discoverability.Adding
KIMI_API_KEYandCORTEX_PROVIDERto the CLI help text ensures users can discover these options via--help.
31-45: Provider detection logic looks good, but consider validating the CORTEX_PROVIDER override.The logic correctly prioritizes
CORTEX_PROVIDERoverride, then falls back to detecting keys in order: OpenAI → Claude → Kimi → Fake. The fallback to'openai'at line 45 is reasonable as a default when no keys or override are found.If
CORTEX_PROVIDERis set to an invalid value (e.g.,"groq"), it will be passed through and fail later inCommandInterpreter. Consider validating the override value against supported providers here to provide earlier, clearer feedback to users.LLM/interpreter.py (2)
51-61: Kimi initialization correctly validates the requests dependency.The try/except for the
requestsimport provides a clear error message if the package is missing. Settingself.client = Nonefor the FAKE provider is appropriate.
157-187: Fake provider implementation is well-designed for testing.The implementation:
- Supports environment-driven commands via
CORTEX_FAKE_COMMANDS- Validates JSON structure properly
- Provides sensible keyword-based defaults for common test cases
- Has a fallback for unknown inputs
This enables deterministic testing without API costs.
videohttps://drive.google.com/file/d/1z9H-PYtC6Gnco3VoG6FnP-1dqDeAno2c/view?usp=sharing updated version. |
There was a problem hiding this comment.
Actionable comments posted: 1
♻️ Duplicate comments (1)
docs/ISSUE_40_KIMI_K2_IMPLEMENTATION.md (1)
25-26: Kimi base URL documented inconsistently across sectionsEarlier sections state the default
KIMI_API_BASE_URLashttps://api.moonshot.ai(and the advanced config example uses that), but the “API Request Format” section useshttps://api.moonshot.cn/v1/chat/completions.Please pick the correct default (ideally whatever
CommandInterpreteractually uses by default) and make all occurrences consistent, or explicitly call out that one is the default and the other is only an example.Also applies to: 48-49, 106-111, 129-145
🧹 Nitpick comments (7)
LLM/test_interpreter.py (3)
49-55: Avoid clearing the entire environment in Kimi init test
@patch.dict(os.environ, {}, clear=True)guarantees a clean slate for Kimi-related env vars, but it also temporarily removes all environment variables for this test, which can cause surprising behavior ifCommandInterpreteror any imported code ever depends on unrelated env settings (e.g.,HOME,PATH). It’s safer to patch only the keys you care about.Consider:
- @patch.dict(os.environ, {}, clear=True) + @patch.dict(os.environ, { + "KIMI_API_KEY": "", + "KIMI_API_BASE_URL": "", + "KIMI_DEFAULT_MODEL": "", + }, clear=False)This keeps non-Kimi env intact while still testing default Kimi behavior.
274-341: Integration tests are well-gated; custom-model env handling now safeThe
TestKimiK2Integrationclass:
- Uses
@unittest.skipUnless(RUN_KIMI_INTEGRATION_TESTS == "1")to keep real Kimi calls out of normal CI.- Validates basic, complex, and validated-command flows against the live API.
- Uses
@patch.dict(os.environ, {'KIMI_DEFAULT_MODEL': 'kimi-k2-0905-preview'})so the custom-model test no longer leaks env state after failures (addressing the earlier cleanup concern).Two minor suggestions:
- Since these are external-network integration tests, consider moving them to a dedicated
integrationmodule/package to keep this unit-test file focused on pure unit tests.- The doc file currently describes the default Kimi model as
kimi-k2, while these tests (and likely the code) usekimi-k2-turbo-previewas the default/custom choice; please align the documentation with the actual default behavior.Both are non-blocking; current structure is functionally correct.
56-73: Kimi success test is solid; asserting request URL and payload structure is a valuable strengthening—Kimi API is stable and OpenAI-compatibleThe success path correctly verifies:
- Parsed commands from
_call_kimi- That
requests.postis called- Presence and value of the
AuthorizationheaderThe suggested assertion for the HTTP contract is well-founded. Kimi's API uses a documented, stable OpenAI-compatible endpoint at
/v1/chat/completionswith standard payload structure (model field, messages array). Adding assertions for the URL path and key JSON fields will reliably catch regressions in_call_kimi's request shape without becoming brittle:call_args, call_kwargs = mock_post.call_args self.assertIn("/v1/chat/completions", call_args[0]) self.assertEqual(call_kwargs["json"]["model"], interpreter.model) self.assertIn("messages", call_kwargs["json"])This strengthens test coverage by locking in the Kimi API contract, which is officially documented and widely implemented by third-party integrators.
docs/ISSUE_40_KIMI_K2_IMPLEMENTATION.md (4)
189-194: Add language hint to fenced code in Testing ResultsThe test output block uses a bare fenced code block. To satisfy markdownlint (MD040) and improve rendering, you can add a language, e.g.:
-``` +```text Ran 143 tests in 10.136s OK (skipped=5)
3-6: Markdownlint nits: bare URLs and minor wordingA few small polish items from static analysis:
- Replace “CLI interface improvements” with just “CLI improvements” to avoid tautology.
- Wrap bare URLs in Markdown links or
<...>to satisfy MD034 (e.g.,[Moonshot AI Platform](https://platform.moonshot.ai/)).These are purely cosmetic and can be deferred.
Also applies to: 83-87, 190-196, 234-236
5-7: Status/Date fields may need updating when PR mergesThe header states
Status: ✅ Implementedand a specific “Date Completed”, but this PR is still open. To avoid confusion for future readers, consider phrasing this as “Planned/Implemented in PR #192 (pending merge)” or updating the status/date when the PR actually lands.
50-55: Consider updatingrequestsversion constraint and documenting versioning approachThe CVE information for
requests>=2.32.4is accurate (all three CVEs are properly fixed at that version). However, the latestrequestsrelease is 2.32.5 (August 2025), making the constraint outdated within months. To make this section age more gracefully, consider:
- Updating to
requests>=2.32.5for the latest patch, or- Noting "(minimum secure version as of [date])" and/or linking to upstream security advisories rather than hard-coding CVE IDs inline.
This ensures the documentation remains correct without requiring constant updates as newer patch versions are released.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (2)
LLM/test_interpreter.py(5 hunks)docs/ISSUE_40_KIMI_K2_IMPLEMENTATION.md(1 hunks)
🧰 Additional context used
🧬 Code graph analysis (1)
LLM/test_interpreter.py (1)
LLM/interpreter.py (3)
APIProvider(9-15)_call_kimi(118-155)parse(227-246)
🪛 LanguageTool
🪛 markdownlint-cli2 (0.18.1)
docs/ISSUE_40_KIMI_K2_IMPLEMENTATION.md
3-3: Bare URL used
(MD034, no-bare-urls)
4-4: Bare URL used
(MD034, no-bare-urls)
190-190: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
234-234: Bare URL used
(MD034, no-bare-urls)
235-235: Bare URL used
(MD034, no-bare-urls)
236-236: Bare URL used
(MD034, no-bare-urls)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: Agent
🔇 Additional comments (3)
LLM/test_interpreter.py (3)
3-6: Sys-module stubbing for OpenAI/Anthropic looks correct and containedUsing
SimpleNamespacepluspatch.dict(sys.modules, ...)insetUpcleanly isolates tests from realopenai/anthropicinstalls and avoidsImportError, andaddCleanup(self.sys_modules_patcher.stop)ensures restoration after each test. This pattern is sound for these unit tests.Also applies to: 15-24
26-31: OpenAI default model assertion aligns with intended behaviorChanging the expected default model to
"gpt-4o"intest_initialization_openaimatches the stated best-practice default elsewhere in the PR and keeps this test tightly coupled to the current interpreter default. No issues here.
74-85: Kimi failure path test matches_call_kimierror behaviorSimulating a
401viaraise_for_status.side_effectand asserting aRuntimeErrorfrom_call_kimicorrectly exercises the generic exception-wrapping logic. This gives good coverage of the negative path without overfitting on the error message.
There was a problem hiding this comment.
Pull request overview
This PR successfully integrates Moonshot AI's Kimi K2 as a new LLM provider for Cortex, expanding multi-LLM capabilities alongside OpenAI and Anthropic. It also introduces a FAKE provider for cost-free testing and adds comprehensive Docker-based integration tests.
Key Changes
- Provider Integration: Added native Kimi K2 support with HTTP-based API calls, configurable via
KIMI_API_KEYenvironment variable - Testing Infrastructure: Introduced FAKE provider for offline testing and Docker-based integration test suite with helper utilities
- Documentation: Added comprehensive testing strategy guide, contribution guidelines, and Kimi K2 implementation reference documentation
Reviewed changes
Copilot reviewed 15 out of 16 changed files in this pull request and generated 16 comments.
Show a summary per file
| File | Description |
|---|---|
test/test_coordinator.py |
Removed obsolete test methods and added unused imports that should be cleaned up |
test/test_cli.py |
Added Kimi and FAKE provider test coverage for API key and provider detection |
test/integration/test_end_to_end.py |
New Docker-based integration tests for end-to-end CLI workflows |
test/integration/docker_utils.py |
Docker container execution utilities for integration testing |
test/integration/__init__.py |
Integration test package initialization |
LLM/test_interpreter.py |
Added Kimi K2 integration tests with real API call validation (conditionally skipped) |
LLM/interpreter.py |
Core Kimi K2 and FAKE provider implementation with API call methods |
cortex/cli.py |
Enhanced provider detection with override support and provider-aware API key handling |
cortex/user_preferences.py |
Major refactoring with improved validation and reduced language support list |
setup.py |
Added requests dependency, updated Python version support (3.8+), corrected license |
requirements.txt |
Added requests and PyYAML dependencies |
requirements-dev.txt |
Added requests to dev dependencies |
LLM/requirements.txt |
Added requests for Kimi HTTP client |
test.md |
Comprehensive testing strategy documentation with Docker requirements |
contribution.md |
New contributor guidelines with coding standards and workflow |
docs/ISSUE_40_KIMI_K2_IMPLEMENTATION.md |
Detailed implementation reference with API examples and configuration |
Comments suppressed due to low confidence (2)
test/test_coordinator.py:2
- Import of 'call' is not used.
from unittest.mock import Mock, patch, call
test/test_coordinator.py:5
- Import of 'time' is not used.
import time
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
There was a problem hiding this comment.
Actionable comments posted: 0
♻️ Duplicate comments (1)
docs/ISSUE_40_KIMI_K2_IMPLEMENTATION.md (1)
21-22: Default Kimi model in documentation doesn't match code.The documentation states
kimi-k2as the default model (lines 21, 48, 138), but the code inLLM/interpreter.pyuseskimi-k2-turbo-previewas the actual default (line 268). This inconsistency may confuse users.Update the documentation to consistently use
kimi-k2-turbo-previewor change the code default ifkimi-k2is the intended value.Also applies to: 48-48, 138-138
🧹 Nitpick comments (9)
docs/ISSUE_40_KIMI_K2_IMPLEMENTATION.md (1)
190-194: Add language specifier to fenced code block.The test results code block is missing a language specifier which aids syntax highlighting and accessibility.
-``` +```text Ran 143 tests in 10.136s OK (skipped=5)</blockquote></details> <details> <summary>cortex/user_preferences.py (3)</summary><blockquote> `328-342`: **Remove unused `key_parts` parameter.** The `key_parts` parameter is never used in `_coerce_value_for_set()`. Either remove it or implement the intended type coercion logic based on the key path. ```diff - def _coerce_value_for_set(self, key_parts: List[str], value: Any) -> Any: + def _coerce_value_for_set(self, value: Any) -> Any: """Coerce string inputs into appropriate types for set()."""And update the call site at line 361:
- coerced = self._coerce_value_for_set(parts, value) + coerced = self._coerce_value_for_set(value)
202-206: Silent exception swallowing may hide configuration issues.The
try-except-passblock silently ignores all load errors during initialization. This could mask configuration file corruption or permission issues that users should be aware of.Consider logging the exception or storing the error state:
try: self.load() except Exception: - # Ignore load errors here; callers can handle later - pass + # Store the error; callers can check or load() will re-raise + self._preferences = UserPreferences()Alternatively, if errors truly should be ignored, at least log them at debug level for troubleshooting.
252-256: Chain exceptions to preserve traceback context.When re-raising exceptions, use
raise ... from eto preserve the original traceback for easier debugging.except yaml.YAMLError as e: - raise ValueError(f"Invalid YAML in config file: {str(e)}") + raise ValueError(f"Invalid YAML in config file: {e!s}") from e except Exception as e: - raise IOError(f"Failed to load config file: {str(e)}") + raise IOError(f"Failed to load config file: {e!s}") from eThe same pattern applies to other exception handlers in
save()(line 286),_create_backup()(line 228), andset()(line 432).test/integration/test_end_to_end.py (1)
107-110: Potential fragile assertion on last output line.The assertion
self.assertIn("OK", result.stdout.splitlines()[-1])will raise anIndexErrorif stdout is empty. Consider adding a guard or using a more robust check.result = self._run("python test/run_all_tests.py", env=env) self.assertTrue(result.succeeded(), msg=result.stderr) - self.assertIn("OK", result.stdout.splitlines()[-1]) + lines = result.stdout.splitlines() + self.assertTrue(lines, "Expected test output but stdout was empty") + self.assertIn("OK", lines[-1])LLM/test_interpreter.py (1)
56-72: Mock response missingraise_for_statusstub.The mock response in
test_call_kimi_successdoesn't stubraise_for_status(). While this works becauseMock()returns another mock by default, explicitly stubbing it ensures the test intent is clear and protects against future changes.mock_response = Mock() mock_response.status_code = 200 + mock_response.raise_for_status = Mock() # Explicitly stub to avoid accidental side effects mock_response.json.return_value = { "choices": [{"message": {"content": '{"commands": ["apt update", "apt install curl"]}'}}] }LLM/interpreter.py (1)
136-137: Redundantrequestsimport inside_call_kimi.The
requestsmodule is already imported (and validated) during_initialize_client()for the KIMI provider (lines 52-58). The import at line 136 is redundant and adds unnecessary overhead on each API call.try: - import requests response = requests.post(Since
self.clientis set to therequestsmodule in_initialize_client(), you can useself.client.post(...)instead for consistency:- import requests - response = requests.post( + response = self.client.post(test/integration/docker_utils.py (2)
24-50:docker_available()is robust; subprocess security lint is low risk hereThe availability check via
docker --versionanddocker infowith small timeouts andshell=Falseis solid for a test helper. Ruff’s S603 warnings here are low risk since arguments are constant and not user-controlled; if those warnings are noisy in CI, you can either# noqa: S603these calls or disable that rule for test modules.
53-106:run_in_dockerhelper looks good; minor f-string / lint tweak and trust boundary noteThe argv-based
dockerinvocation plusbash -lcinside the container is appropriate for an integration-test harness and avoids host-side shell injection; just ensure callers only pass trustedcommandstrings since they execute with full shell semantics in the container. To address the RUF010 hint and avoidstr()inside an f-string, you can simplify the volume mount construction:- for host_path, container_path in mounts or []: - docker_cmd.extend([ - "-v", - f"{str(host_path.resolve())}:{container_path}", - ]) + for host_path, container_path in mounts or []: + docker_cmd.extend( + [ + "-v", + f"{host_path.resolve()!s}:{container_path}", + ] + )
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (7)
LLM/interpreter.py(9 hunks)LLM/test_interpreter.py(5 hunks)cortex/user_preferences.py(5 hunks)docs/ISSUE_40_KIMI_K2_IMPLEMENTATION.md(1 hunks)test/integration/docker_utils.py(1 hunks)test/integration/test_end_to_end.py(1 hunks)test/test_coordinator.py(1 hunks)
🚧 Files skipped from review as they are similar to previous changes (1)
- test/test_coordinator.py
🧰 Additional context used
🧬 Code graph analysis (1)
LLM/test_interpreter.py (1)
LLM/interpreter.py (3)
APIProvider(9-15)_call_kimi(118-153)parse(225-244)
🪛 LanguageTool
🪛 markdownlint-cli2 (0.18.1)
docs/ISSUE_40_KIMI_K2_IMPLEMENTATION.md
3-3: Bare URL used
(MD034, no-bare-urls)
4-4: Bare URL used
(MD034, no-bare-urls)
190-190: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
234-234: Bare URL used
(MD034, no-bare-urls)
235-235: Bare URL used
(MD034, no-bare-urls)
236-236: Bare URL used
(MD034, no-bare-urls)
🪛 Ruff (0.14.7)
LLM/interpreter.py
55-55: Avoid specifying long messages outside the exception class
(TRY003)
147-147: Abstract raise to an inner function
(TRY301)
147-147: Avoid specifying long messages outside the exception class
(TRY003)
150-150: Abstract raise to an inner function
(TRY301)
150-150: Avoid specifying long messages outside the exception class
(TRY003)
153-153: Avoid specifying long messages outside the exception class
(TRY003)
153-153: Use explicit conversion flag
Replace with conversion flag
(RUF010)
163-163: Avoid specifying long messages outside the exception class
(TRY003)
165-165: Avoid specifying long messages outside the exception class
(TRY003)
228-228: Avoid specifying long messages outside the exception class
(TRY003)
test/integration/docker_utils.py
32-32: subprocess call: check for execution of untrusted input
(S603)
40-40: subprocess call: check for execution of untrusted input
(S603)
48-48: Consider moving this statement to an else block
(TRY300)
89-89: Use explicit conversion flag
Replace with conversion flag
(RUF010)
97-97: subprocess call: check for execution of untrusted input
(S603)
cortex/user_preferences.py
142-142: Do not catch blind exception: Exception
(BLE001)
152-152: Do not catch blind exception: Exception
(BLE001)
204-206: try-except-pass detected, consider logging the exception
(S110)
204-204: Do not catch blind exception: Exception
(BLE001)
226-226: Consider moving this statement to an else block
(TRY300)
227-227: Do not catch blind exception: Exception
(BLE001)
228-228: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
228-228: Avoid specifying long messages outside the exception class
(TRY003)
228-228: Use explicit conversion flag
Replace with conversion flag
(RUF010)
250-250: Consider moving this statement to an else block
(TRY300)
253-253: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
253-253: Avoid specifying long messages outside the exception class
(TRY003)
253-253: Use explicit conversion flag
Replace with conversion flag
(RUF010)
254-254: Do not catch blind exception: Exception
(BLE001)
255-255: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
255-255: Avoid specifying long messages outside the exception class
(TRY003)
255-255: Use explicit conversion flag
Replace with conversion flag
(RUF010)
283-283: Consider moving this statement to an else block
(TRY300)
285-285: Do not catch blind exception: Exception
(BLE001)
286-286: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
286-286: Avoid specifying long messages outside the exception class
(TRY003)
286-286: Use explicit conversion flag
Replace with conversion flag
(RUF010)
328-328: Unused method argument: key_parts
(ARG002)
370-370: Do not catch blind exception: Exception
(BLE001)
371-371: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
371-371: Avoid specifying long messages outside the exception class
(TRY003)
375-375: Abstract raise to an inner function
(TRY301)
375-375: Avoid specifying long messages outside the exception class
(TRY003)
377-377: Abstract raise to an inner function
(TRY301)
377-377: Avoid specifying long messages outside the exception class
(TRY003)
382-382: Abstract raise to an inner function
(TRY301)
382-382: Avoid specifying long messages outside the exception class
(TRY003)
385-385: Abstract raise to an inner function
(TRY301)
385-385: Avoid specifying long messages outside the exception class
(TRY003)
387-387: Abstract raise to an inner function
(TRY301)
387-387: Avoid specifying long messages outside the exception class
(TRY003)
392-392: Abstract raise to an inner function
(TRY301)
392-392: Avoid specifying long messages outside the exception class
(TRY003)
399-399: Do not catch blind exception: Exception
(BLE001)
400-400: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
400-400: Avoid specifying long messages outside the exception class
(TRY003)
403-403: Abstract raise to an inner function
(TRY301)
403-403: Avoid specifying long messages outside the exception class
(TRY003)
410-410: Abstract raise to an inner function
(TRY301)
410-410: Avoid specifying long messages outside the exception class
(TRY003)
412-412: Abstract raise to an inner function
(TRY301)
412-412: Avoid specifying long messages outside the exception class
(TRY003)
417-417: Abstract raise to an inner function
(TRY301)
417-417: Avoid specifying long messages outside the exception class
(TRY003)
419-419: Abstract raise to an inner function
(TRY301)
419-419: Avoid specifying long messages outside the exception class
(TRY003)
426-426: Abstract raise to an inner function
(TRY301)
426-426: Avoid specifying long messages outside the exception class
(TRY003)
430-430: Consider moving this statement to an else block
(TRY300)
432-432: Within an except clause, raise exceptions with raise ... from err or raise ... from None to distinguish them from errors in exception handling
(B904)
432-432: Avoid specifying long messages outside the exception class
(TRY003)
432-432: Use explicit conversion flag
Replace with conversion flag
(RUF010)
457-457: Avoid specifying long messages outside the exception class
(TRY003)
539-539: Avoid specifying long messages outside the exception class
(TRY003)
🔇 Additional comments (5)
test/integration/test_end_to_end.py (1)
27-37: Add return type annotation to_runmethod.The method has parameter type hints but is missing the return type annotation.
- def _run(self, command: str, env: dict | None = None) -> DockerRunResult: + def _run(self, command: str, env: dict | None = None) -> DockerRunResult:The signature already includes
-> DockerRunResultso this appears to be addressed.LLM/test_interpreter.py (1)
274-285: LGTM - Integration test class is well-guarded.The skip decorator with
RUN_KIMI_INTEGRATION_TESTSenvironment variable check and thesetUpmethod'sskipTestfor missing API key provide appropriate guards for real API tests. The docstring correctly shows the unittest command.LLM/interpreter.py (2)
155-166: LGTM -_call_fakecorrectly returns commands directly.The implementation properly validates that
commandsis a list and returns it directly, avoiding the redundant JSON parsing that was previously flagged.
260-269: LGTM - Default model resolution is well-implemented.The
_default_modelmethod correctly provides current model defaults for each provider, with the Kimi provider supporting environment variable override viaKIMI_DEFAULT_MODEL. The OpenAI default has been updated togpt-4oaddressing the previous deprecation concern.test/integration/docker_utils.py (1)
11-21:DockerRunResultdataclass is a clean abstractionThe dataclass and
succeeded()helper neatly wrap subprocess results; no changes needed.
There was a problem hiding this comment.
Actionable comments posted: 1
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
LLM/interpreter.py (1)
30-34: Use_default_model()in__init__to fix incorrect defaults for KIMI/FAKERight now,
__init__hard-codes default models:if model: self.model = model else: self.model = "gpt-4" if self.provider == APIProvider.OPENAI else "claude-sonnet-4-20250514"You’ve introduced
_default_model()with provider-specific defaults (including KIMI and FAKE), but__init__doesn’t call it. As a result:
- KIMI and FAKE instances created without an explicit
modelwill incorrectly get"claude-sonnet-4-20250514"(theelsebranch), never using the KIMI/FAKE defaults defined in_default_model().- There are now two divergent sources of truth for default models, which is brittle.
This is a functional bug for KIMI/FAKE defaults and a maintainability issue overall. Suggest centralizing on
_default_model():def __init__( self, api_key: str, provider: str = "openai", model: Optional[str] = None ): self.api_key = api_key self.provider = APIProvider(provider.lower()) - - if model: - self.model = model - else: - self.model = "gpt-4" if self.provider == APIProvider.OPENAI else "claude-sonnet-4-20250514" - + + # Use provider-specific defaults, overridden by explicit `model` if provided. + self.model = model or self._default_model() + self._initialize_client()With this, KIMI and FAKE will correctly pick up
"kimi-k2-turbo-preview"(orKIMI_DEFAULT_MODEL) and"fake-local-model"when no explicit model is passed, and there’s a single place to adjust defaults going forward.Also applies to: 260-269
🧹 Nitpick comments (1)
LLM/interpreter.py (1)
37-62: Unify KIMI client setup and HTTP call; avoid re-importingrequestsFor the KIMI path,
_initialize_client()importsrequests, assigns it toself.client, and sets_kimi_base_url, but_call_kimi()re-importsrequestsand doesn’t useself.client. This duplication slightly complicates testability and makes the client setup pattern inconsistent with the other providers.You can simplify and make this more consistent by relying on
self.clientin_call_kimi()and dropping the second import:def _call_kimi(self, user_input: str) -> List[str]: """Call the Kimi K2 HTTP API and parse the response body.""" @@ - try: - import requests - response = requests.post( + try: + response = self.client.post( f"{self._kimi_base_url.rstrip('/')}/v1/chat/completions", headers=headers, json=payload, timeout=60, )(Optionally, you could make
self.clientarequests.Session()in_initialize_client()if you want connection pooling, but that’s not required for correctness.)Also, please double-check that the endpoint path (
/v1/chat/completions) and response shape (choices[0].message.content) still match the current Kimi K2 API.Also applies to: 118-154
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
LLM/interpreter.py(9 hunks)
🧰 Additional context used
🧬 Code graph analysis (1)
LLM/interpreter.py (1)
cortex/user_preferences.py (2)
get(288-315)validate(463-500)
🪛 Ruff (0.14.7)
LLM/interpreter.py
55-55: Avoid specifying long messages outside the exception class
(TRY003)
147-147: Abstract raise to an inner function
(TRY301)
147-147: Avoid specifying long messages outside the exception class
(TRY003)
150-150: Abstract raise to an inner function
(TRY301)
150-150: Avoid specifying long messages outside the exception class
(TRY003)
153-153: Avoid specifying long messages outside the exception class
(TRY003)
153-153: Use explicit conversion flag
Replace with conversion flag
(RUF010)
163-163: Avoid specifying long messages outside the exception class
(TRY003)
165-165: Avoid specifying long messages outside the exception class
(TRY003)
228-228: Avoid specifying long messages outside the exception class
(TRY003)
🔇 Additional comments (2)
LLM/interpreter.py (2)
10-15: Enum additions for KIMI/FAKE look consistent with the rest of the interpreterThe new
APIProvider.KIMIandAPIProvider.FAKEmembers align with the branches in_initialize_clientandparse(), and the string values match theproviderargument usage. No changes needed here.
64-80: Docstrings and high-level flow for parsing/validation are clearThe added/updated docstrings around
_get_system_prompt, provider-specific call methods,_parse_commands,_validate_commands,parse, andparse_with_contextaccurately describe the behavior and make the NL→commands pipeline much easier to follow. No issues here.Also applies to: 82-98, 101-116, 187-203, 205-223, 225-244, 246-258
|
I saw both of your video. There is not clear showcasing of kimi usage. Have you even use api key to test if your code is working. Create a separate video just showcase kimi usage or how you tested on your system. Testing as in not running test files fyi, proper usage. |
|
I can pull up echo command to show API is in environment, or I can add show short function in CLI to show which LLM is getting used for command execution. |
|
Echo is more like print statement. |
|
@dhvll Could you review this PR? Would appreciate your testing perspective on the Kimi K2 provider. Thanks! |
There was a problem hiding this comment.
Actionable comments posted: 0
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
LLM/interpreter.py (1)
31-40: Missing default model assignment for KIMI and FAKE providers.When
modelis not explicitly provided andprovideris"kimi"or"fake",self.modelis never assigned. This causes anAttributeErrorin_call_kimi(line 136) when building the payload.Apply this diff to add default models for the new providers:
if model: self.model = model else: if self.provider == APIProvider.OPENAI: self.model = "gpt-4" elif self.provider == APIProvider.CLAUDE: self.model = "claude-sonnet-4-20250514" + elif self.provider == APIProvider.KIMI: + self.model = os.environ.get("KIMI_DEFAULT_MODEL", "kimi-k2-turbo-preview") + elif self.provider == APIProvider.FAKE: + self.model = "fake-local-model" elif self.provider == APIProvider.OLLAMA: self.model = "llama3.2" # Default Ollama modelAlternatively, consolidate this logic by calling
self._default_model()(defined at line 306) instead of duplicating defaults.
♻️ Duplicate comments (1)
LLM/interpreter.py (1)
169-176: Missing key existence check before accessingdata["commands"].If the JSON in
CORTEX_FAKE_COMMANDSis valid but lacks the"commands"key, line 174 raises a bareKeyErrorinstead of a descriptiveValueError.This was flagged in a previous review. Add an explicit key check:
try: data = json.loads(payload) except json.JSONDecodeError as exc: raise ValueError("CORTEX_FAKE_COMMANDS must contain valid JSON") from exc + if "commands" not in data: + raise ValueError("CORTEX_FAKE_COMMANDS must contain a 'commands' key") if not isinstance(data["commands"], list): raise ValueError("'commands' must be a list in CORTEX_FAKE_COMMANDS") return data["commands"]
🧹 Nitpick comments (2)
LLM/interpreter.py (2)
145-152: Redundantrequestsimport; use the stored client instead.
requestsis already imported and stored inself.clientduring_initialize_client()(line 63). The re-import here is unnecessary.try: - import requests - response = requests.post( + response = self.client.post( f"{self._kimi_base_url.rstrip('/')}/v1/chat/completions", headers=headers, json=payload, timeout=60, )
306-315:_default_model()is defined but never called.This method duplicates the default model logic already present in
__init__(lines 34-39) but is never invoked. This creates maintenance burden and potential drift between the two locations.Either:
- Remove
_default_model()if it was added speculatively, or- Refactor
__init__to use it:if model: self.model = model else: - if self.provider == APIProvider.OPENAI: - self.model = "gpt-4" - elif self.provider == APIProvider.CLAUDE: - self.model = "claude-sonnet-4-20250514" - elif self.provider == APIProvider.OLLAMA: - self.model = "llama3.2" # Default Ollama model + self.model = self._default_model()Note: This requires moving the method definition before
__init__or deferring the call until after full initialization.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (3)
LLM/interpreter.py(10 hunks)cortex/cli.py(3 hunks)requirements.txt(1 hunks)
🚧 Files skipped from review as they are similar to previous changes (1)
- requirements.txt
🧰 Additional context used
🧬 Code graph analysis (1)
cortex/cli.py (1)
cortex/branding.py (1)
cx_print(55-75)
🪛 Ruff (0.14.7)
LLM/interpreter.py
61-61: Avoid specifying long messages outside the exception class
(TRY003)
157-157: Abstract raise to an inner function
(TRY301)
157-157: Avoid specifying long messages outside the exception class
(TRY003)
160-160: Abstract raise to an inner function
(TRY301)
160-160: Avoid specifying long messages outside the exception class
(TRY003)
163-163: Avoid specifying long messages outside the exception class
(TRY003)
163-163: Use explicit conversion flag
Replace with conversion flag
(RUF010)
173-173: Avoid specifying long messages outside the exception class
(TRY003)
175-175: Avoid specifying long messages outside the exception class
(TRY003)
272-272: Avoid specifying long messages outside the exception class
(TRY003)
🔇 Additional comments (7)
LLM/interpreter.py (3)
57-64: LGTM for KIMI client initialization.The import check and base URL configuration are properly handled. The fallback to
https://api.moonshot.aiis reasonable.
178-195: LGTM for fake provider defaults.The safe defaults use only
echocommands, making them safe for dry-run and integration tests. The keyword matching and fallback logic are reasonable.
278-281: LGTM for new provider routing.The KIMI and FAKE providers are correctly routed to their respective handler methods, consistent with the existing pattern.
cortex/cli.py (4)
68-71: Auto-detection offakeprovider viaCORTEX_FAKE_COMMANDSpresence.This implicit detection allows tests to set
CORTEX_FAKE_COMMANDSwithout needingCORTEX_PROVIDER=fake. While convenient for testing, it could surprise users who set the variable for other reasons (e.g., debugging) without intending to switch providers.Consider documenting this behavior or requiring explicit
CORTEX_PROVIDER=fakefor production safety.
76-99: LGTM for provider-aware API key retrieval.The method correctly maps providers to their environment variables and provides helpful guidance when keys are missing. The Ollama placeholder and structured error messaging are appropriate.
137-143: LGTM for fake provider handling in install flow.The fake provider correctly bypasses real API key requirements with a sensible default, enabling offline testing without additional configuration.
830-831: LGTM for CLI help documentation.The epilog correctly documents the new
KIMI_API_KEYandCORTEX_PROVIDERenvironment variables, including all supported provider options.
There was a problem hiding this comment.
Actionable comments posted: 1
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
requirements-dev.txt(1 hunks)
🔇 Additional comments (1)
requirements-dev.txt (1)
4-5: New dev dependencies appropriately support test infrastructure.The addition of
pytest-mock,requests,PyYAML, andpre-commitaligns well with the PR's integration test suite and multi-provider support (Kimi K2, FAKE). These dependencies enable mocking, HTTP interactions, YAML configuration handling, and pre-commit consistency checks.Also applies to: 8-8, 25-25
| PyYAML>=6.0.0 | ||
|
|
||
| # Code Quality | ||
| black>=23.0.0 |
There was a problem hiding this comment.
Remove duplicate black entry with conflicting versions.
The file contains two entries for the black package with different version constraints (line 11: >=23.0.0 and line 22: >=24.0.0). This creates ambiguous dependency resolution and will cause issues with package managers.
Apply this diff to remove the duplicate:
# Code Quality
- black>=23.0.0
pylint>=2.17.0
mypy>=1.0.0Retain line 22's black>=24.0.0 as the single, authoritative version constraint.
Also applies to: 22-22
🤖 Prompt for AI Agents
In requirements-dev.txt around lines 11 to 22, there are duplicate entries for
black with conflicting versions; remove the earlier entry at line 11
(`black>=23.0.0`) so only the authoritative `black>=24.0.0` at line 22 remains,
ensuring no duplicate/conflicting constraints in the file.
|
3 participants
Summary
Testing
this pull request has test file which was need for issue #40 but it fixed #16 because of that as well.
Summary by CodeRabbit
New Features
Refactor
Documentation
Tests
Chores
✏️ Tip: You can customize this high-level summary in your review settings.