fix(ci): set ai_fallback_model to satisfy pr-reviewer-action v2.1.10

[Tanguille]

Summary

Supersedes/absorbs #4329 (the Renovate bump to v2.1.10) by also fixing the validation break it introduces.

v2.1.10 changelog: "Make fallback base_url/api_format/api_key inherit from primary" — ai_fallback_base_url now silently inherits from ai_base_url when unset. The same release also validates that ai_fallback_model is set whenever ai_fallback_base_url is set (even inherited), and fails before making any model call otherwise.

Fix

• Bump the pinned digest to v2.1.10.
• Set ai_fallback_model: qwen-3.6-fast — a real, distinct second attempt via the self-hosted model, not a duplicate of ai_model. litellm's own routerSettings.fallbacks already retries omniroute failures internally with qwen-3.6-fast, so this action-level fallback only fires if that whole internal chain fails too.

Evidence

Run 30878353953 (self-review of the v2.1.10 bump): ERROR: AI_FALLBACK_MODEL is required when AI_FALLBACK_BASE_URL is set, duration_ms=42 — fails before any model call.

Test plan

• Next PR review run completes without the AI_FALLBACK_MODEL is required error

Summary by CodeRabbit

• Chores
  • Updated the automated pull request review workflow.
  • Configured a fallback model to improve review processing reliability.
  • Updated the review automation action to its latest version.

[deepsource-io]

DeepSource Code Review

We reviewed changes in ddca87e...aafaa41 on this pull request. Below is the summary for the review, and you can see the individual issues we found as inline review comments.

See full review on DeepSource ↗

PR Report Card

Overall Grade

Security   Reliability   Complexity   Hygiene

Code Review Summary

Analyzer	Status	Updated (UTC)	Details
JavaScript		Aug 4, 2026 2:07p.m.	Review ↗
Shell		Aug 4, 2026 2:07p.m.	Review ↗

---

Important

AI Review is run only on demand for your team. We're only showing results of static analysis review right now. To trigger AI Review, comment @deepsourcebot review on this thread.

[coderabbitai]

📝 Walkthrough

Walkthrough

The PR updates misospace/pr-reviewer-action to v2.1.10 and adds the required ai_fallback_model: qwen-3.6-fast setting to the PR review workflow.

Changes

PR reviewer workflow

Layer / File(s)	Summary
Update reviewer action and fallback configuration .github/workflows/agent-pr-review.yaml	The workflow updates the action from v2.1.9 to v2.1.10 and configures qwen-3.6-fast as the fallback model. Added comments documenting inherited fallback URL validation and fallback behavior.

Estimated code review effort: 1 (Trivial) | ~2 minutes

Possibly related PRs

• Tanguille/cluster#4329: Adds the required fallback-model parameter for the v2.1.10 workflow update.
• Tanguille/cluster#4309: Configures AI fallback models and related behavior in the same workflow.
• Tanguille/cluster#4311: Updates the same reviewer action in the same workflow.

Suggested labels: type/patch, renovate/github-action

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly describes the main change: setting ai_fallback_model to satisfy pr-reviewer-action v2.1.10 validation.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/pr-reviewer-fallback-model

---

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

AI Automated Review

Analysis engine: omniroute@http://litellm.ai.svc.cluster.local/v1 (openai)

Recommendation

Approve. This is a small, well-documented CI workflow fix that bumps misospace/pr-reviewer-action from v2.1.9 to v2.1.10 and adds the ai_fallback_model input required by the new version's validation. The change is consistent with the upstream release notes, the repository's conventions, and the PR's own evidence of the failure mode it fixes.

Change-by-Change Findings

1. Digest bump: misospace/pr-reviewer-action@1cb6bed… → @6f4fb72… (v2.1.9 → v2.1.10)

Verified against the upstream release API. The new digest corresponds to tag v2.1.10. The bump spans exactly one release (v2.1.9 → v2.1.10), so no intermediate releases need enumeration.

2. Added ai_fallback_model: qwen-3.6-fast

The PR body explains that v2.1.10 makes ai_fallback_base_url inherit from ai_base_url when unset, and then requires ai_fallback_model whenever ai_fallback_base_url is set (even inherited), failing fast before any model call otherwise. The PR provides concrete evidence of this failure (ERROR: AI_FALLBACK_MODEL is required when AI_FALLBACK_BASE_URL is set, duration_ms=42).

The chosen value qwen-3.6-fast is consistent with the existing comment in the same file describing litellm's routerSettings.fallbacks chain, and the explanatory comments correctly note that this is a distinct second attempt rather than a no-op retry of ai_model. Since ai_fallback_api_key and ai_fallback_api_format also inherit from primary per the v2.1.10 release notes, no additional fallback config is needed.

Release Notes

Upstream release notes for misospace/pr-reviewer-action v2.1.10 (fetched via gh_api):

• Make fallback base_url/api_format/api_key inherit from primary (PR 448) — this is the behavioral change that drives this PR; ai_fallback_base_url now silently inherits from ai_base_url when unset, and the action now validates that ai_fallback_model is set whenever ai_fallback_base_url is set.
• fix(routing): reduce smart escalation — routing behavior adjustment.
• fix(routing): ignore environmental unknowns — routing behavior adjustment.

No breaking changes or security fixes are called out in the release notes. The prior release (v2.1.9) added Forgejo Authorized Integration authentication support, which is not relevant to this repository's GitHub-based workflow.

Standards Compliance

• Conventional Commit title: fix(ci): set ai_fallback_model to satisfy pr-reviewer-action v2.1.10 matches the fix(scope): description format required by AGENTS.md. ✓
• Small, focused change: Single file modified, 9 additions / 1 deletion. ✓
• No shell scripts touched: shellcheck requirement not applicable. ✓
• No Kubernetes manifests touched: flate/kustomize validation not applicable. ✓
• Comments explain rationale: The added comments document the v2.1.10 behavior change and why qwen-3.6-fast is the correct fallback model, following the repository's pattern of thorough inline documentation. ✓

Unknowns or Needs Verification

• Tool harness findings are pending — the corpus reports "Tool harness planning pending," so no tool-harness validation output was available for this review. This is a CI workflow change with no Kubernetes manifests, so the impact of missing harness output is minimal.
• The upstream action.yml validation logic was not fully verified — the fetched action.yml content was base64-encoded and truncated, so the exact validation code path could not be inspected directly. However, the release notes ("Make fallback base_url/api_format/api_key inherit from primary") and the PR body's concrete failure evidence are mutually consistent and support the described behavior.
• The digest-to-tag correspondence was verified via the release API (tag v2.1.10 exists at the referenced commit), but the commit object itself was not fetched to independently confirm the SHA matches the tag. The release API response confirms the tag exists; this is a standard Renovate-style pinned digest bump.

Test Plan

• The PR's test plan ("Next PR review run completes without the AI_FALLBACK_MODEL is required error") is appropriate — the fix is validated by the next actual workflow run. No further verification is required before merge.

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In @.github/workflows/agent-pr-review.yaml:
- Line 36: Update the configuration using the misospace/pr-reviewer-action
reference so the action’s fallback model is distinct from the primary model and
does not retry the same URL/model combination; preserve the existing LiteLLM
routing configuration while selecting an appropriate alternate fallback model.

🪄 Autofix

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 6f6a93b6-d6ef-46c0-8fd4-c5b8abcb1723

📥 Commits

Reviewing files that changed from the base of the PR and between ddca87e and 19e4dce.

📒 Files selected for processing (1)

• .github/workflows/agent-pr-review.yaml

🔗 Linked repositories identified

CodeRabbit considers these linked repositories for cross-repo context during reviews:

• Tanguille/LLMKube (auto-detected)
• Tanguille/toolhive (auto-detected)
• Tanguille/pr-reviewer-action (auto-detected)
• Tanguille/renovate-presets (auto-detected)
• Tanguille/litellm-operator (auto-detected)
• Tanguille/2x-R9700-RDNA4-GFX1201-sglang-inference (auto-detected)