From 88608fb12bdc67def93e8beaea489527d9ae5469 Mon Sep 17 00:00:00 2001 From: Jory Irving Date: Sun, 14 Jun 2026 22:14:32 -0600 Subject: [PATCH] chore(ai-review): adopt pr-reviewer v1.3.0 native_loop + org-var tokens MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit - Bump the action pin v1.2.9 → v1.3.0 (3479bf5). - Switch tool_mode plan_execute_loop → native_loop (the v1.3.0 flagship multi-hop tool-calling harness; requires v1.3.0 for streaming + the in-conversation verdict). - Wire ai_max_tokens to an org-wide variable: ${{ vars.AI_MAX_TOKENS || '16000' }} so it can be tuned once across the misospace org (16000 fallback until set). - Raise the per-turn loop budget for the thinking model: tool_planning_max_tokens 1500 → 16000 (1500 truncates a reasoning model every loop turn), and tool_planning_timeout_sec 45 → 300 for longer streamed turns; add tool_loop_wall_clock_sec 300 (default 120 is short for the slow self-hosted reviewer). Co-Authored-By: Claude Opus 4.8 (1M context) --- .github/workflows/ai-pr-review.yaml | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) diff --git a/.github/workflows/ai-pr-review.yaml b/.github/workflows/ai-pr-review.yaml index 67f7411..08f39e0 100644 --- a/.github/workflows/ai-pr-review.yaml +++ b/.github/workflows/ai-pr-review.yaml @@ -39,7 +39,7 @@ jobs: - name: Review PR with reusable AI reviewer if: github.event_name == 'pull_request' id: review - uses: misospace/pr-reviewer-action@4126a2701a5e9adcf7c48ac6eb678e2e3a2bb2a0 # v1.2.9 + uses: misospace/pr-reviewer-action@3479bf51ab9133414ae07af783bff985d3929884 # v1.3.0 with: github_token: ${{ steps.app-token.outputs.token }} ai_primary_retries: "3" @@ -49,6 +49,7 @@ jobs: ai_model: ${{ vars.PRIMARY_MODEL }} ai_api_key: ${{ secrets.LITELLM_API_KEY }} ai_response_format: json_object + ai_max_tokens: ${{ vars.AI_MAX_TOKENS || '16000' }} ai_fallback_base_url: ${{ vars.LITELLM_URL }} ai_fallback_api_format: ${{ vars.FALLBACK_FORMAT }} ai_fallback_model: ${{ vars.FALLBACK_MODEL }} @@ -61,12 +62,13 @@ jobs: context_limit_mode: normal ci_status_check: "true" ci_timeout_sec: "600" - tool_mode: plan_execute_loop + tool_mode: native_loop tool_max_rounds: "2" tool_max_requests: "4" - tool_planning_timeout_sec: "45" + tool_loop_wall_clock_sec: "300" + tool_planning_timeout_sec: "300" tool_planning_max_context_bytes: "15000" - tool_planning_max_tokens: "1500" + tool_planning_max_tokens: "16000" tool_max_response_bytes: "12000" tool_allowed_gh_api_repos: "*" tool_request_timeout_sec: "15"