Merge remote-tracking branch 'upstream/main' into feat/kenari-provider

# Conflicts:
#	webview-ui/src/components/settings/__tests__/ApiOptions.spec.tsx

Author: doedja

.changeset/fix-anthropic-custom-model-id-fallback.md

@@ -0,0 +1,11 @@
+---
+"zoo-code": patch
+---
+
+Fix Anthropic provider silently replacing a custom/unrecognized `apiModelId` with the hardcoded default model.
+
+`AnthropicHandler.getModel()` coerced any `apiModelId` not present in the static `anthropicModels` table down to `anthropicDefaultModelId` ("claude-sonnet-4-5"), and that coerced id was what actually got sent as `model` in the API request -- silently ignoring a user-configured custom model name (e.g. a custom Anthropic-compatible deployment or proxy). This produced confusing "model does not exist" errors for the default model instead of the model the user actually selected (#418).
+
+The same fallback also affected capability lookups used to build the `thinking` request parameter: an unrecognized id fell back to the default model's info, which can be from an older model generation with a different API contract, causing the request to use the legacy `thinking: {type: "enabled", budget_tokens}` shape and get rejected with a 400 by models that require `{type: "adaptive"}`.
+
+The model id sent to the API now always honors a user-configured `apiModelId`. For unrecognized values, capabilities are best-effort guessed by matching known model-family substrings (mirroring the existing `BedrockHandler.guessModelInfoFromId` heuristic) instead of defaulting to `anthropicDefaultModelId`'s info.

.github/workflows/codeql.yml

@@ -52,7 +52,7 @@ jobs:
 
     # Initializes the CodeQL tools for scanning.
     - name: Initialize CodeQL
-      uses: github/codeql-action/init@dd903d2e4f5405488e5ef1422510ee31c8b32357 # v3
+      uses: github/codeql-action/init@411c4c9a36b3fca4d674f06b6396b2c6d23522c6 # v3
       with:
         languages: ${{ matrix.language }}
         build-mode: ${{ matrix.build-mode }}
@@ -80,6 +80,6 @@ jobs:
         exit 1
 
     - name: Perform CodeQL Analysis
-      uses: github/codeql-action/analyze@dd903d2e4f5405488e5ef1422510ee31c8b32357 # v3
+      uses: github/codeql-action/analyze@411c4c9a36b3fca4d674f06b6396b2c6d23522c6 # v3
       with:
         category: "/language:${{matrix.language}}"

.github/workflows/label-pr-review-state.yml

@@ -26,7 +26,7 @@ jobs:
         with:
           script: |
             const { owner, repo } = context.repo;
-            const stateLabels = ['awaiting-author', 'awaiting-review'];
+            const stateLabels = ['awaiting-author', 'awaiting-review', 'has-conflicts'];
 
             // When triggered by a single PR event, only reconcile that PR.
             // The hourly schedule and workflow_dispatch reconcile all open PRs.
@@ -125,6 +125,22 @@ jobs:
                   continue;
                 }
 
+                // `mergeable`/`mergeable_state` are only returned by the single-PR GET
+                // endpoint, and are computed asynchronously by GitHub — a PR fetched via
+                // pulls.list (schedule/workflow_dispatch runs) never has them, and even a
+                // single-PR fetch can return `null`/"unknown" if the merge check hasn't
+                // finished yet. Re-fetch the single PR to get a fresh value, and treat
+                // "unknown" as not-yet-computed rather than as conflicting.
+                const prDetail = prNumber
+                  ? pr
+                  : (await github.rest.pulls.get({ owner, repo, pull_number: pr.number })).data;
+
+                if (prDetail.mergeable === false && prDetail.mergeable_state === 'dirty') {
+                  core.info(`PR #${pr.number}: has merge conflicts — labeling has-conflicts`);
+                  await reconcileLabels(pr, 'has-conflicts');
+                  continue;
+                }
+
                 // Check CI status for required checks on the PR's head commit only.
                 // Scoping to required checks avoids advisory checks (e.g. codecov/patch)
                 // incorrectly blocking label assignment on otherwise-ready PRs.
@@ -137,9 +153,28 @@ jobs:
                   }),
                 ]);
 
+                // listForRef returns every check run ever recorded on the ref, including
+                // stale superseded ones (e.g. a failed run later re-run green). Branch
+                // protection and the PR UI only consider the latest run per check name, so
+                // reduce to that before evaluating — otherwise a single stale failure makes
+                // ciFailed true forever and state labels never come back. See issue #884.
+                //
+                // Unlike listReviews (which documents oldest-first order), listForRef's
+                // ordering is unspecified, so we pick the latest by run.id — GitHub assigns
+                // monotonically increasing IDs, and id is never null (a freshly re-queued
+                // run can have started_at: null, which would lose a string comparison
+                // against an older completed run's timestamp).
+                const latestByName = new Map();
+                for (const run of checkRuns) {
+                  const prev = latestByName.get(run.name);
+                  if (!prev || run.id > prev.id) {
+                    latestByName.set(run.name, run);
+                  }
+                }
+
                 // Filter to required checks only (or all checks if rules unavailable).
                 // Always exclude this workflow's own run to avoid self-referential loops.
-                const relevantRuns = checkRuns.filter(run => {
+                const relevantRuns = [...latestByName.values()].filter(run => {
                   if (run.name === 'Reconcile PR review state labels') return false;
                   return requiredCheckNames ? requiredCheckNames.has(run.name) : true;
                 });

.github/workflows/nightly-publish.yml

@@ -41,7 +41,21 @@ jobs:
             exit 1
           fi
 
+      - name: Skip if this push is a release merge
+        id: release-check
+        run: |
+          commit_subject=$(git log -1 --format=%s)
+          skip=false
+
+          if [[ "$commit_subject" =~ ^chore:\ prepare\ v[0-9]+\.[0-9]+\.[0-9]+\ release ]]; then
+            echo "Commit '${commit_subject}' looks like a release-prep merge; skipping nightly pre-release."
+            skip=true
+          fi
+
+          echo "skip=${skip}" >> "$GITHUB_OUTPUT"
+
       - name: Set pre-release version
+        if: steps.release-check.outputs.skip != 'true'
         id: version
         env:
           RUN_NUMBER: ${{ github.run_number }}
@@ -61,13 +75,15 @@ jobs:
           EOF
 
       - name: Build workspace packages
+        if: steps.release-check.outputs.skip != 'true'
         env:
           PKG_RELEASE_CHANNEL: prerelease
         run: |
           pnpm --filter @roo-code/build build
           pnpm --filter @roo-code/vscode-webview build
 
       - name: Package pre-release VSIX
+        if: steps.release-check.outputs.skip != 'true'
         env:
           POSTHOG_API_KEY: ${{ secrets.POSTHOG_API_KEY }}
           PKG_RELEASE_CHANNEL: prerelease
@@ -76,6 +92,7 @@ jobs:
           pnpm --filter ./src exec vsce package --pre-release --no-dependencies --out ../bin
 
       - name: Verify VSIX contents
+        if: steps.release-check.outputs.skip != 'true'
         env:
           VERSION_NUMBER: ${{ steps.version.outputs.number }}
         run: |
@@ -87,6 +104,7 @@ jobs:
           grep -q "extension/webview-ui/audio/celebration.wav" /tmp/zoo-code-vsix-contents.txt
 
       - name: Validate packaged manifest identity
+        if: steps.release-check.outputs.skip != 'true'
         env:
           VERSION_NUMBER: ${{ steps.version.outputs.number }}
         run: |
@@ -98,12 +116,36 @@ jobs:
           test "$artifact_name" = "zoo-code"
           test "$artifact_publisher" = "ZooCodeOrganization"
 
-      # Open VSX is intentionally excluded: it has no pre-release channel concept,
-      # so pre-release builds would surface as the latest stable version for all users.
       - name: Publish pre-release to VS Code Marketplace
+        if: steps.release-check.outputs.skip != 'true'
         env:
           VSCE_PAT: ${{ secrets.VSCE_PAT }}
           VERSION_NUMBER: ${{ steps.version.outputs.number }}
         run: |
-          npx @vscode/vsce publish --pre-release --packagePath "bin/zoo-code-${VERSION_NUMBER}.vsix"
-          echo "Published ZooCodeOrganization.zoo-code ${VERSION_NUMBER} as a VS Code Marketplace pre-release"
+          npx @vscode/vsce publish --pre-release --skip-duplicate --packagePath "bin/zoo-code-${VERSION_NUMBER}.vsix"
+          echo "Published or skipped existing ZooCodeOrganization.zoo-code ${VERSION_NUMBER} as a VS Code Marketplace pre-release"
+
+      # The VSIX built above with `vsce package --pre-release` already carries the
+      # Microsoft.VisualStudio.Code.PreRelease manifest property, which is what Open
+      # VSX reads to flag the version. `ovsx publish` ignores --pre-release for an
+      # already-packaged .vsix (it only applies when ovsx does the packaging itself),
+      # so it's intentionally omitted here.
+      #
+      # Open VSX's "latest" alias resolves to the highest semver across stable and
+      # pre-release alike (pre-release only breaks ties at equal major.minor.patch),
+      # so a nightly build can transiently become "latest" until the next stable
+      # release outranks it. This mirrors how Marketplace pre-release users already
+      # track the newest published version, so it's an accepted trade-off here too.
+      - name: Publish pre-release to Open VSX Registry
+        if: steps.release-check.outputs.skip != 'true'
+        env:
+          OVSX_PAT: ${{ secrets.OVSX_PAT }}
+          VERSION_NUMBER: ${{ steps.version.outputs.number }}
+        run: |
+          set -o pipefail
+          publish_output=$(pnpm exec ovsx publish "bin/zoo-code-${VERSION_NUMBER}.vsix" --skip-duplicate 2>&1 | tee /dev/stderr)
+          if echo "$publish_output" | grep -q "is already published"; then
+            echo "ZooCodeOrganization.zoo-code ${VERSION_NUMBER} was already published to Open VSX; skipped."
+          else
+            echo "Published ZooCodeOrganization.zoo-code ${VERSION_NUMBER} as an Open VSX pre-release"
+          fi

CHANGELOG.md

@@ -1,5 +1,24 @@
 # Zoo Code Changelog
 
+## [3.68.0]
+
+### Minor Changes
+
+- Add Friendli provider with GLM-5.2 support for another hosted way to use the latest GLM model (#722 by @Lee-Si-Yoon, PR #721 by @Lee-Si-Yoon)
+- Add native thinking/reasoning support for Ollama models to preserve reasoning output end-to-end (#831 by @navedmerchant, PR #832 by @navedmerchant)
+- Fix(anthropic): honor custom `apiModelId` selections instead of silently defaulting to `claude-sonnet-4-5` (#418 by @tatianadenel-devops, #843 by @grizmin, PR #842 by @grizmin)
+- Fix(ollama): correctly handle tool results and prevent premature context condensing (#847 by @navedmerchant, PR #848 by @navedmerchant)
+- Improve Anthropic Vertex Claude content block handling for more reliable responses (#788 by @daewoongoh, PR #789 by @daewoongoh)
+- Fix(task-lifecycle): preserve the parent-child link when a delegated subtask is interrupted (#560 by @edelauna, PR #787 by @edelauna)
+- Refactor: remove the deprecated `openai-error-handler` shim and use the shared `error-handler` directly (#766 by @daewoongoh, PR #767 by @daewoongoh)
+- Feat(nightly-publish): publish Open VSX pre-releases and skip nightly publish on release merges (#784 by @edelauna, PR #790 by @edelauna)
+- Fix(ci): don't skip fork-PR label reconciliation on scheduled and manual runs (PR #234 by @app/roomote)
+- Fix(label-pr-review-state): detect merge conflicts and label PRs with `has-conflicts` (PR #269 by @app/roomote)
+- Chore(deps): update the `github/codeql-action` digest to `411c4c9` (PR #803 by @app/renovate)
+- Chore(deps): update `@types/react` to `v18.3.31` (PR #805 by @app/renovate)
+- Chore(deps): update `axios` to `v1.18.1` (PR #806 by @app/renovate)
+- Chore: merge the v3.66.0 release preparation branch into `main` (PR #795 by @navedmerchant)
+
 ## [3.66.0]
 
 ### Minor Changes

README.md

@@ -53,22 +53,15 @@
 You can find a quick guide for migrating from Roo Code to Zoo Code in the [Roo→Zoo migration guide](https://docs.zoocode.dev/roo-to-zoo-migration). We plan to try and help users as they transition over, we have our [Reddit](https://www.reddit.com/r/ZooCode) and [Discord](https://discord.gg/VxfP4Vx3gX)
 for this exact support, so if you are having problems or if you have question, jump on and ask.
 
-## What's New in v3.66.0
-
-- **Claude Sonnet 5 support** — the latest Claude model is now available across Anthropic, Bedrock, and Vertex providers
-- **Semble v0.4.1 upgrade** — flattened result parsing and localized status messages
-- **Task-lifecycle status transition guard** — a new status transition guard and startup delegation reconciliation prevent invalid task state transitions
-- Fix: LiteLLM cache key collision and silent fallback to a non-existent default model
-- Fix: reliable auto context condensing for the VS Code Language Model API
-- Fix: ThinkingBudget now supports `xhigh` and all extended reasoning effort values
-- Fix: round-trip DeepSeek `reasoning_content` in thinking mode to prevent 400 errors
-- Fix: base64-encode Gemini `thoughtSignature` bypass token to fix the Vertex AI empty-response loop
-- Fix: provider cache reset after settings import
-- Fix: atomically serialize `reopenParentFromDelegation`
-- Fix: shell default profile name type guard
-- Security: dependency-review, invisible-char detection, and least-privilege workflow permissions
-- Upgrade `@anthropic-ai/sdk` to 0.104.1 and `@anthropic-ai/vertex-sdk` to 0.17.1
-- Dependency and tooling updates
+## What's New in v3.68.0
+
+- **Friendli provider with GLM-5.2 support** — use the latest GLM model through Friendli.
+- **Native Ollama thinking/reasoning support** — preserve reasoning output end-to-end when you use Ollama models.
+- **Anthropic custom `apiModelId` fix** — custom Anthropic model IDs now stay selected instead of silently falling back to `claude-sonnet-4-5`.
+- Fix: Ollama provider tool result handling and premature context condensing.
+- Fix: preserve the parent-child task link when a delegated subtask is interrupted.
+- Improve Anthropic Vertex Claude content block handling for more reliable responses.
+- CI, nightly publishing, and dependency/tooling updates.
 
 <details>
   <summary>🌐 Available languages</summary>

apps/cli/package.json

@@ -41,7 +41,7 @@
 		"@roo-code/config-eslint": "workspace:^",
 		"@roo-code/config-typescript": "workspace:^",
 		"@types/node": "20.19.43",
-		"@types/react": "18.3.23",
+		"@types/react": "18.3.31",
 		"@vitest/coverage-v8": "4.1.9",
 		"ink-testing-library": "4.0.0",
 		"rimraf": "6.0.1",

apps/cli/src/ui/components/autocomplete/triggers/HistoryTrigger.tsx

@@ -21,7 +21,7 @@ export interface HistoryResult extends AutocompleteItem {
 	/** Mode the task was run in */
 	mode?: string
 	/** Task status */
-	status?: "active" | "completed" | "delegated"
+	status?: "active" | "completed" | "delegated" | "interrupted"
 }
 
 /**
@@ -133,8 +133,22 @@ export function createHistoryTrigger(config: HistoryTriggerConfig): Autocomplete
 
 		renderItem: (item: HistoryResult, isSelected: boolean) => {
 			// Status indicator
-			const statusIcon = item.status === "completed" ? "✓" : item.status === "active" ? "●" : "○"
-			const statusColor = item.status === "completed" ? "green" : item.status === "active" ? "yellow" : "gray"
+			const statusIcon =
+				item.status === "completed"
+					? "✓"
+					: item.status === "active"
+						? "●"
+						: item.status === "interrupted"
+							? "⏸"
+							: "○"
+			const statusColor =
+				item.status === "completed"
+					? "green"
+					: item.status === "active"
+						? "yellow"
+						: item.status === "interrupted"
+							? "cyan"
+							: "gray"
 
 			// Mode indicator (if available)
 			const modeText = item.mode ? ` [${item.mode}]` : ""
@@ -178,7 +192,7 @@ export function toHistoryResult(item: {
 	totalCost?: number
 	workspace?: string
 	mode?: string
-	status?: "active" | "completed" | "delegated"
+	status?: "active" | "completed" | "delegated" | "interrupted"
 }): HistoryResult {
 	return {
 		key: item.id, // Use task ID as the unique key

apps/cli/src/ui/components/autocomplete/triggers/__tests__/HistoryTrigger.test.tsx

@@ -188,6 +188,24 @@ describe("HistoryTrigger", () => {
 			expect(output).toContain("○")
 		})
 
+		it("should render interrupted status with correct indicator", () => {
+			const trigger = createHistoryTrigger({ getHistory: () => mockHistoryItems })
+
+			const interruptedItem: HistoryResult = {
+				key: "task-interrupted",
+				id: "task-interrupted",
+				task: "Interrupted subtask waiting to resume",
+				ts: Date.now() - 1000 * 60 * 5,
+				mode: "ask",
+				status: "interrupted",
+			}
+			const { lastFrame } = render(trigger.renderItem(interruptedItem, false) as React.ReactElement)
+
+			const output = lastFrame()
+			// Should contain the interrupted status indicator (⏸)
+			expect(output).toContain("⏸")
+		})
+
 		it("should render selected items with different styling", () => {
 			const trigger = createHistoryTrigger({ getHistory: () => mockHistoryItems })
 

apps/cli/src/ui/types.ts

@@ -109,7 +109,7 @@ export interface TaskHistoryItem {
 	totalCost?: number
 	workspace?: string
 	mode?: string
-	status?: "active" | "completed" | "delegated"
+	status?: "active" | "completed" | "delegated" | "interrupted"
 	tokensIn?: number
 	tokensOut?: number
 }