vybestack
diff --git a/‎.gitattributes‎
Lines changed: 6 additions & 0 deletions b/‎.gitattributes‎
Lines changed: 6 additions & 0 deletions
diff --git a/‎.github/workflows/ci.yml‎
Lines changed: 20 additions & 2 deletions b/‎.github/workflows/ci.yml‎
Lines changed: 20 additions & 2 deletions
diff --git a/‎.github/workflows/luther.yml‎
Lines changed: 2 additions & 1 deletion b/‎.github/workflows/luther.yml‎
Lines changed: 2 additions & 1 deletion
diff --git a/‎.github/workflows/pr-review.yml‎
Lines changed: 30 additions & 15 deletions b/‎.github/workflows/pr-review.yml‎
Lines changed: 30 additions & 15 deletions
diff --git a/‎.github/workflows/release.yml‎
Lines changed: 7 additions & 1 deletion b/‎.github/workflows/release.yml‎
Lines changed: 7 additions & 1 deletion
diff --git a/‎dev-docs/integration-tests.md‎
Lines changed: 19 additions & 0 deletions b/‎dev-docs/integration-tests.md‎
Lines changed: 19 additions & 0 deletions
diff --git a/‎docs/cli/authentication.md‎
Lines changed: 0 additions & 1 deletion b/‎docs/cli/authentication.md‎
Lines changed: 0 additions & 1 deletion
@@ -22,3 +22,9 @@ Makefile eol=lf
 *.eot binary
 *.ttf binary
 *.otf binary
+
+# Avoid huge diffs for autogenerated/lock files in review prompts
+package-lock.json -diff
+packages/ui/bun.lock -diff
+docs/cli/configuration.md -diff
+schemas/settings.schema.json -diff
@@ -167,7 +167,12 @@ jobs:
 
       - name: 'Install UI dependencies (bun)'
         run: |-
-          cd packages/ui && bun install
+          cd packages/ui
+          if [ "${{ runner.os }}" = "Windows" ]; then
+            bun install --no-save
+          else
+            bun install
+          fi
 
       - name: Fix rollup platform dependency
         run: |
@@ -380,7 +385,12 @@ jobs:
 
       - name: 'Install UI dependencies (bun)'
         run: |-
-          cd packages/ui && bun install
+          cd packages/ui
+          if [ "${{ runner.os }}" = "Windows" ]; then
+            bun install --no-save
+          else
+            bun install
+          fi
 
       - name: 'Build project'
         run: |-
@@ -401,6 +411,14 @@ jobs:
           fi
         shell: 'bash'
 
+      - name: 'Limit Vitest threads on Windows'
+        if: runner.os == 'Windows'
+        run: |
+          echo "VITEST_MAX_THREADS=1" >> "$GITHUB_ENV"
+          echo "VITEST_MIN_THREADS=1" >> "$GITHUB_ENV"
+          echo "VITEST_TEST_TIMEOUT=30000" >> "$GITHUB_ENV"
+          echo "VITEST_POOL_TIMEOUT=60000" >> "$GITHUB_ENV"
+
       - name: 'Run tests and generate reports'
         env:
           # Provider configuration from repository secrets/variables
 
@@ -483,6 +483,7 @@ jobs:
 
             echo "Running LLxprt for attempt ${attempt}"
             llxprt_log="${attempt_dir}/llxprt.log"
+            context_limit="${LLXPRT_CONTEXT_LIMIT:-200000}"
             set +e
             cat "$prompt_file" | llxprt \
               --provider "${LLXPRT_DEFAULT_PROVIDER}" \
@@ -491,7 +492,7 @@ jobs:
               --key "${OPENAI_API_KEY}" \
               --set modelparam.temperature=1 \
               --set modelparam.max_tokens=10000 \
-              --set context-limit=121000 \
+              --set context-limit="${context_limit}" \
               --set base-url="${OPENAI_BASE_URL}" \
               --set shell-replacement=true 2>&1 | tee "$llxprt_log"
             llxprt_exit="${PIPESTATUS[1]}"
 
@@ -40,6 +40,7 @@ jobs:
       OPENAI_BASE_URL: '${{ vars.OPENAI_BASE_URL }}'
       LLXPRT_DEFAULT_MODEL: '${{ vars.LLXPRT_DEFAULT_MODEL }}'
       LLXPRT_DEFAULT_PROVIDER: '${{ vars.LLXPRT_DEFAULT_PROVIDER }}'
+      LLXPRT_CONTEXT_LIMIT: "${{ vars.LLXPRT_CONTEXT_LIMIT || '200000' }}"
       LLXPRT_DEBUG: "${{ vars.DEBUG_NAMESPACES || 'llxprt:*' }}"
       DEBUG_OUTPUT: 'stderr'
     steps:
@@ -127,7 +128,7 @@ jobs:
           if [[ ! -s "${issues_file}" ]]; then
             {
               echo "<!-- llxprt-pr-review -->"
-              echo "## ⚠️ LLxprt PR Review blocked"
+              echo "## LLxprt PR Review blocked"
               echo
               echo "- No linked issues were detected in this PR's description."
               echo "- Please reference an existing issue with text such as \`Fixes #123\` so the automated review knows what problem to evaluate."
@@ -237,8 +238,8 @@ jobs:
           git diff --stat "${BASE_SHA}" "${PR_HEAD_SHA}" > review/diffstat.txt
           git diff --name-status "${BASE_SHA}" "${PR_HEAD_SHA}" > review/changed-files.txt
           git diff --numstat "${BASE_SHA}" "${PR_HEAD_SHA}" > review/numstat.txt
-          git diff -U3 "${BASE_SHA}" "${PR_HEAD_SHA}" > review/diff.patch
-          head -n 4000 review/diff.patch > review/diff-truncated.patch
+          git diff -U0 "${BASE_SHA}" "${PR_HEAD_SHA}" > review/diff.patch
+          head -n 1000 review/diff.patch > review/diff-truncated.patch
           grep -Ei '(/tests?/|__tests__|\\.spec\\.|\\.test\\.)' review/changed-files.txt > review/test-files.txt || true
 
       - name: 'Build LLxprt prompt'
@@ -278,28 +279,35 @@ jobs:
             "review/coverage-latest.txt",
             "No coverage summary comment was found on this PR yet.",
           );
-          const diffstat = readIfExists("review/diffstat.txt", "");
-          const filesList = readIfExists("review/changed-files.txt", "");
+          const diffstat = process.env.DIFF_STAT_ENABLED === "true"
+            ? readIfExists("review/diffstat.txt", "")
+            : "";
+          const filesList = process.env.DIFF_FILES_ENABLED === "true"
+            ? readIfExists("review/changed-files.txt", "")
+            : "";
           const testFiles = readIfExists(
             "review/test-files.txt",
             "No explicit test files were touched.",
           );
-          const diffPatch = readIfExists("review/diff-truncated.patch", "");
+          const diffPatch = process.env.DIFF_PATCH_ENABLED === "true"
+            ? readIfExists(
+                "review/diff-truncated.patch",
+                "Patch omitted to keep prompt within context limit.",
+              )
+            : "";
 
           const issueSection = issues.length
             ? issues
                 .map((issue) => {
-                  const summary = clean(issue.body || "", 800);
+                  const summary = clean(issue.body || "", 120);
                   return `- #${issue.number} ${issue.title}
-            - URL: ${issue.url}
-            - State: ${issue.state}
             - Summary: ${summary}`;
                 })
                 .join("\n")
             : "No issues were expanded (this should not happen).";
 
           const prBody =
-            clean(pr.body || "", 2000) || "No PR description provided.";
+            clean(pr.body || "", 400) || "No PR description provided.";
 
           const prompt = `You are LLxprt, an autonomous code reviewer for the LLxprt Code repository.
           Focus strictly on code introduced in this pull request. Ignore pre-existing code unless it directly interacts with the changes shown.
@@ -333,8 +341,8 @@ jobs:
           ## Test-focused files touched
           ${testFiles}
 
-          ## Patch (first 4000 lines)
-          ${diffPatch}
+          ## Patch (first 1000 lines)
+          ${diffPatch || "Patch omitted to keep prompt within context limit."}
 
           ## Doc-only change
           ${process.env.DOCS_ONLY || 'unknown'}
@@ -370,7 +378,7 @@ jobs:
           - Additional bullets on test gaps or strengths (mention if tests feel like “mock theater”).
 
           **Verdict**
-          - Default to \`⚠️ Needs Work\` unless you can cite specific evidence (code + tests) that risk is low; only start with \`✅ Ready\` when requirements are fully satisfied, the change is truly documentation-only, or adequate automated coverage exists.
+          - Default to Needs Work unless you can cite specific evidence (code + tests) that risk is low; only start with Ready when requirements are fully satisfied, the change is truly documentation-only, or adequate automated coverage exists.
 
           If information is missing, state assumptions explicitly. Keep the overall length under 350 words.
           `;
@@ -381,10 +389,17 @@ jobs:
       - name: 'Run LLxprt review'
         if: steps.issue_gate.outputs.should_review == 'true'
         id: 'llxprt'
+        env:
+          DIFF_FILES_ENABLED: 'true'
+          DIFF_STAT_ENABLED: 'true'
+          DIFF_PATCH_ENABLED: 'false'
         run: |
           set -euo pipefail
           prompt_file="review/prompt.md"
           llxprt_log="review/llxprt.log"
+          context_limit="${LLXPRT_CONTEXT_LIMIT:-200000}"
+          echo "LLXPRT_CONTEXT_LIMIT raw: '${LLXPRT_CONTEXT_LIMIT-<unset>}'" >&2
+          echo "LLXPRT_CONTEXT_LIMIT evaluated: '${context_limit}'" >&2
           set +e
           cat "${prompt_file}" | llxprt \
             --provider "${LLXPRT_DEFAULT_PROVIDER}" \
@@ -393,15 +408,15 @@ jobs:
             --key "${OPENAI_API_KEY}" \
             --set modelparam.temperature=1 \
             --set modelparam.max_tokens=10000 \
-            --set context-limit=121000 \
+            --set context-limit="${context_limit}" \
             --set base-url="${OPENAI_BASE_URL}" \
             --set shell-replacement=true | tee "${llxprt_log}"
           llxprt_status=${PIPESTATUS[1]}
           set -e
           if [[ ${llxprt_status} -ne 0 ]]; then
             {
               echo "<!-- llxprt-pr-review -->"
-              echo "## ⚠️ LLxprt PR Review infrastructure failure"
+              echo "## WARNING: LLxprt PR Review infrastructure failure"
               echo
               echo "The automated reviewer failed with exit code ${llxprt_status}. Please inspect the workflow logs (LLxprt section) and re-run once resolved."
             } > review/comment.md
 
@@ -150,7 +150,13 @@ jobs:
         run: npm ci
 
       - name: Install UI dependencies (bun)
-        run: cd packages/ui && bun install
+        run: |
+          cd packages/ui
+          if [ "${{ runner.os }}" = "Windows" ]; then
+            bun install --no-save
+          else
+            bun install
+          fi
 
       - name: Get the version
         id: version
 
@@ -44,6 +44,25 @@ To run a single test by its name, use the `--test-name-pattern` flag:
 npm run test:e2e -- --test-name-pattern "reads a file"
 ```
 
+### Regenerating model responses
+
+Some integration tests use faked out model responses, which may need to be
+regenerated from time to time as the implementations change.
+
+To regenerate these golden files, set the REGENERATE_MODEL_GOLDENS environment
+variable to "true" when running the tests, for example:
+
+**WARNING**: If running locally you should review these updated responses for
+any information about yourself or your system that gemini may have included in
+these responses.
+
+```bash
+REGENERATE_MODEL_GOLDENS="true" npm run test:e2e
+```
+
+**WARNING**: Make sure you run **await rig.cleanup()** at the end of your test,
+else the golden files will not be updated.
+
 ### Deflaking a test
 
 Before adding a **new** integration test, you should test it at least 5 times with the deflake script to make sure that it is not flaky.
 
@@ -19,7 +19,6 @@ LLxprt Code supports multiple authentication methods for various AI providers. F
       ```bash
       export GOOGLE_CLOUD_PROJECT="YOUR_PROJECT_ID"
       ```
-
       - For repeated use, you can add the environment variable to your [.env file](#persisting-environment-variables-with-env-files) or your shell's configuration file (like `~/.bashrc`, `~/.zshrc`, or `~/.profile`). For example, the following command adds the environment variable to a `~/.bashrc` file:
 
       ```bash