Merged
Conversation
Reviewer's GuideIntroduces a local MCP server and Cursor configuration for QuantLab, extends the desktop app tooling, documents agent/MCP usage, and surfaces validation/source artifact links in the pre‑trade UI while tightening a related test expectation. Sequence diagram for quantLab MCP tool invocationssequenceDiagram
actor Agent
participant Cursor
participant MCP_server_quantlab_local as quantlab_local_server
participant PythonCLI as Python_main_py
participant NodeSmoke as Node_smoke_script
participant FileSystem
Agent->>Cursor: Trigger MCP tool (e.g. quantlab_check)
Cursor->>MCP_server_quantlab_local: JSON RPC tool call over stdio
activate MCP_server_quantlab_local
alt Python_based_tool
MCP_server_quantlab_local->>PythonCLI: spawn python main.py with flags
activate PythonCLI
PythonCLI-->>MCP_server_quantlab_local: exitCode, stdout, stderr
deactivate PythonCLI
else Desktop_smoke_tool
MCP_server_quantlab_local->>NodeSmoke: spawn node scripts/smoke.js
activate NodeSmoke
NodeSmoke-->>MCP_server_quantlab_local: exitCode, stdout, stderr
deactivate NodeSmoke
else Read_file_tool
MCP_server_quantlab_local->>FileSystem: Resolve and read relative_path
FileSystem-->>MCP_server_quantlab_local: File contents or error
end
MCP_server_quantlab_local-->>Cursor: Formatted text content (truncated if needed)
deactivate MCP_server_quantlab_local
Cursor-->>Agent: Display tool result in IDE
File-Level Changes
Tips and commandsInteracting with Sourcery
Customizing Your ExperienceAccess your dashboard to:
Getting Help
|
![sourcery-ai[bot]](https://avatars.githubusercontent.com/in/48477?s=60&v=4){kind=small}
There was a problem hiding this comment.
Hey - I've found 3 issues, and left some high level feedback:
- In
mcp-server.mjs, thequantlab_read_filetool’sinputSchemashould be a Zod object schema (e.g.,z.object({ relative_path: z.string().describe(...) })) rather than a plain object literal to match the MCP SDK’s expectations and ensure proper validation. - When formatting command lines in
formatProcessResult(e.g., forquantlab_check,quantlab_version, etc.), consider using thePYTHON_EXECUTABLEconstant instead of hardcodingpythonso that the reported command matches the actual executable used, especially whenQUANTLAB_PYTHONis set.
Prompt for AI Agents
Please address the comments from this code review:
## Overall Comments
- In `mcp-server.mjs`, the `quantlab_read_file` tool’s `inputSchema` should be a Zod object schema (e.g., `z.object({ relative_path: z.string().describe(...) })`) rather than a plain object literal to match the MCP SDK’s expectations and ensure proper validation.
- When formatting command lines in `formatProcessResult` (e.g., for `quantlab_check`, `quantlab_version`, etc.), consider using the `PYTHON_EXECUTABLE` constant instead of hardcoding `python` so that the reported command matches the actual executable used, especially when `QUANTLAB_PYTHON` is set.
## Individual Comments
### Comment 1
<location path="desktop/mcp-server.mjs" line_range="146-151" />
<code_context>
+ return formatProcessResult("quantlab_desktop_smoke", result, "node scripts/smoke.js");
+ });
+
+ server.registerTool("quantlab_read_file", {
+ description: "Read a text file within the QuantLab repository.",
+ inputSchema: {
+ relative_path: z.string().describe("Path relative to the QuantLab repository root"),
+ },
+ }, async ({ relative_path }) => {
+ const resolvedPath = path.resolve(PROJECT_ROOT, relative_path);
+ const relative = path.relative(PROJECT_ROOT, resolvedPath);
</code_context>
<issue_to_address>
**issue (bug_risk):** `inputSchema` for `quantlab_read_file` is not a Zod schema object, which may break validation and tool metadata.
Here `inputSchema` is a plain object instead of a Zod schema, so the MCP SDK may not treat it as a valid schema (breaking validation or metadata). This should be wrapped in `z.object(...)`, e.g.
```ts
inputSchema: z.object({
relative_path: z
.string()
.describe("Path relative to the QuantLab repository root"),
}),
```
and the handler signature adjusted if the SDK now passes an object matching that schema shape.
</issue_to_address>
### Comment 2
<location path="desktop/mcp-server.mjs" line_range="107-108" />
<code_context>
+ server.registerTool("quantlab_check", {
+ description: "Run the standard QuantLab health check.",
+ }, async () => {
+ const result = await runPythonCli(["--check"], 120000);
+ return formatProcessResult("quantlab_check", result, "python main.py --check");
+ });
+
</code_context>
<issue_to_address>
**nitpick:** The command string passed to `formatProcessResult` is hardcoded to `python` and can diverge from `PYTHON_EXECUTABLE`.
Because the display string is hardcoded as `"python main.py --check"` while the actual executable comes from `PYTHON_EXECUTABLE`, the shown command can differ from what was run (e.g. when `QUANTLAB_PYTHON` points to another interpreter). To keep them aligned, construct the display command from `PYTHON_EXECUTABLE`, e.g.:
```js
const commandLine = `${PYTHON_EXECUTABLE} main.py --check`;
const result = await runPythonCli(["--check"], 120000);
return formatProcessResult("quantlab_check", result, commandLine);
```
You can apply the same approach to the other `quantlab_*` tools.
</issue_to_address>
### Comment 3
<location path="test/test_research_ui_server.py" line_range="514-515" />
<code_context>
assert payload["handoff_id"] == "handoff-newer"
assert payload["latest_validation_path"] == str(newer)
assert payload["latest_validation_href"] == "/outputs/pretrade_handoff/newer/pretrade_handoff_validation.json"
+ assert payload["source_artifact_path"] == "C:\\Users\\marce\\Documents\\meta_trade\\tests\\fixtures\\expected_quantlab_handoff.json"
+ assert payload["source_artifact_href"] is None
</code_context>
<issue_to_address>
**issue (testing):** Avoid hard-coded, machine-specific absolute path in test assertion
`source_artifact_path` is asserted against a Windows-specific absolute path (with a username and local directory layout), which will fail on other machines/CI and on non-Windows platforms. Please build the expected path from the same test/fixture base (e.g., `tmp_path`, `Path`, or a fixture helper) and compare against that, or relax the check (e.g., `endswith`/`in`) to keep the test portable.
</issue_to_address>Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.
Comment on lines
+146
to
+151
| server.registerTool("quantlab_read_file", { | ||
| description: "Read a text file within the QuantLab repository.", | ||
| inputSchema: { | ||
| relative_path: z.string().describe("Path relative to the QuantLab repository root"), | ||
| }, | ||
| }, async ({ relative_path }) => { |
There was a problem hiding this comment.
issue (bug_risk): inputSchema for quantlab_read_file is not a Zod schema object, which may break validation and tool metadata.
Here inputSchema is a plain object instead of a Zod schema, so the MCP SDK may not treat it as a valid schema (breaking validation or metadata). This should be wrapped in z.object(...), e.g.
inputSchema: z.object({
relative_path: z
.string()
.describe("Path relative to the QuantLab repository root"),
}),and the handler signature adjusted if the SDK now passes an object matching that schema shape.
Comment on lines
+107
to
+108
| const result = await runPythonCli(["--check"], 120000); | ||
| return formatProcessResult("quantlab_check", result, "python main.py --check"); |
There was a problem hiding this comment.
nitpick: The command string passed to formatProcessResult is hardcoded to python and can diverge from PYTHON_EXECUTABLE.
Because the display string is hardcoded as "python main.py --check" while the actual executable comes from PYTHON_EXECUTABLE, the shown command can differ from what was run (e.g. when QUANTLAB_PYTHON points to another interpreter). To keep them aligned, construct the display command from PYTHON_EXECUTABLE, e.g.:
const commandLine = `${PYTHON_EXECUTABLE} main.py --check`;
const result = await runPythonCli(["--check"], 120000);
return formatProcessResult("quantlab_check", result, commandLine);You can apply the same approach to the other quantlab_* tools.
Comment on lines
+514
to
+515
| assert payload["source_artifact_path"] == "C:\\Users\\marce\\Documents\\meta_trade\\tests\\fixtures\\expected_quantlab_handoff.json" | ||
| assert payload["source_artifact_href"] is None |
There was a problem hiding this comment.
issue (testing): Avoid hard-coded, machine-specific absolute path in test assertion
source_artifact_path is asserted against a Windows-specific absolute path (with a username and local directory layout), which will fail on other machines/CI and on non-Windows platforms. Please build the expected path from the same test/fixture base (e.g., tmp_path, Path, or a fixture helper) and compare against that, or relax the check (e.g., endswith/in) to keep the test portable.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds project-wide Cursor rules, a local MCP server for QuantLab validation tools, and repo guidance so agents use the same operating context.
Validation:
Summary by Sourcery
Add a local MCP server and agent guidance, and surface QuantLab validation artifacts in the research UI pretrade panel.
New Features:
Enhancements:
Build:
Documentation: