package.json (1)

12-14: Duplicated devDependencies — these already exist in apps/server/package.json.

In a Yarn/npm workspaces monorepo, test dependencies that are only consumed by apps/server should live exclusively in that workspace's package.json. Having them duplicated at the root can cause version drift and confusion about which package actually depends on them. Consider removing them here.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@package.json` around lines 12 - 14, Remove the duplicated test
devDependencies from the root package.json: delete the "@types/supertest" and
"supertest" entries under "devDependencies" in this file and keep those
dependencies declared only in apps/server/package.json, ensuring the workspace
relies on the workspace-level declaration to avoid version drift and ambiguity.

apps/server/lib/github.ts (2)

122-154: any[] types in Issue interface reduce type safety.

labels and milestone are typed as any[]/any | null. If the shape is known (even partially), consider at least a minimal interface (e.g., { id: number; name: string } for labels) to catch misuse at compile time. Fine to defer if these fields are unused today.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/server/lib/github.ts` around lines 122 - 154, The Issue interface
currently weakens type safety by using any[] for labels and any | null for
milestone; define small, explicit types (e.g., Label with id:number and
name:string, optional color/string fields; Milestone with id:number,
title:string, optional description/string and due_on/date) and replace labels:
any[] with labels: Label[] and milestone: any | null with milestone: Milestone |
null in the Issue interface (update any related usages/imports and adjust types
in functions that construct or read Issue.labels and Issue.milestone such as
where Issue objects are parsed or returned).

---

192-243: Consider adding a max-page safeguard to fetchPaginated.

There's no upper bound on the number of pages fetched. A repo with many thousands of comments could cause this loop to make hundreds of sequential API calls, risking rate-limit exhaustion and long request latency. A simple maxPages parameter (with a sensible default, e.g., 50) would prevent runaway pagination.

Proposed fix

-async function fetchPaginated<T>(url: string, token: string): Promise<T[]> {
+async function fetchPaginated<T>(url: string, token: string, maxPages: number = 50): Promise<T[]> {
   let results: T[] = [];
   let nextUrl: string | null = url;
+  let page = 0;
 
-  while (nextUrl) {
+  while (nextUrl && page < maxPages) {
+    page++;
     const response = await fetch(nextUrl, {

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/server/lib/github.ts` around lines 192 - 243, Add a bounded pagination
safeguard to fetchPaginated by adding an optional maxPages parameter (number)
with a sensible default (e.g., 50) to the function signature and tracking a
pagesFetched counter inside the while (nextUrl) loop; increment pagesFetched
each iteration and break or throw a clear error once pagesFetched >= maxPages to
avoid runaway requests, ensuring you still return accumulated results or surface
a descriptive error if the limit is hit. Reference: function fetchPaginated,
variables nextUrl and results, introduce pagesFetched and maxPages.

apps/server/app.ts (1)

483-500: Parallelize independent getIssueComments and getIssue calls.

getIssueComments and getIssue are independent API calls but are executed sequentially, doubling latency for the "all comments" view. Use Promise.all to fetch both concurrently.

Proposed fix

   try {
     const client = new GitHubClient(token);
-    let comments = await client.getIssueComments(owner, repo, issueNum);
-    let issue: Issue | null = null;
 
     if (commentNum !== null) {
+      let comments = await client.getIssueComments(owner, repo, issueNum);
       if (commentNum < 1 || commentNum > comments.length) {
         const errorMarkdown = `# Error 404\n\nComment not found.\n\n**Comment Number:** ${commentNum}\n**Issue:** #${issueNum}\n**Total Comments:** ${comments.length}\n\nComment number must be between 1 and ${comments.length}.`;
         return res.status(404)
           .set("Content-Type", "text/markdown; charset=utf-8")
           .send(errorMarkdown);
       }
       comments = [comments[commentNum - 1]!];
+      const markdown = formatIssueAsMarkdown(owner, repo, issueNum, null, comments);
+      return res.status(200)
+        .set("Content-Type", "text/markdown; charset=utf-8")
+        .send(markdown);
     } else {
-        // Fetch issue details only if not fetching a specific comment (or maybe fetch it anyway?)
-        // Fetching it anyway to include in the full view
-        issue = await client.getIssue(owner, repo, issueNum);
+      const [comments, issue] = await Promise.all([
+        client.getIssueComments(owner, repo, issueNum),
+        client.getIssue(owner, repo, issueNum),
+      ]);
+      const markdown = formatIssueAsMarkdown(owner, repo, issueNum, issue, comments);
+      return res.status(200)
+        .set("Content-Type", "text/markdown; charset=utf-8")
+        .send(markdown);
     }
-
-    const markdown = formatIssueAsMarkdown(owner, repo, issueNum, issue, comments);
-
-    return res.status(200)
-      .set("Content-Type", "text/markdown; charset=utf-8")
-      .send(markdown);

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@apps/server/app.ts` around lines 483 - 500, The code currently calls
client.getIssueComments(...) then awaits client.getIssue(...) sequentially when
commentNum is null; change it to run them in parallel using Promise.all to
reduce latency: for the branch where commentNum === null, call
Promise.all([client.getIssueComments(owner, repo, issueNum),
client.getIssue(owner, repo, issueNum)]) and destructure the results into
comments and issue respectively; keep the existing behavior for the commentNum
!== null branch (only fetch comments), and ensure the surrounding try/catch and
variable names (GitHubClient, getIssueComments, getIssue, comments, issue,
commentNum) are preserved.