Fix PR AI review for MIMO gateway

[utopiafar]

What changed

• Pin PR AI Review Claude Code runs to mimo-v2.5-pro for the configured MIMO Anthropic-compatible gateway.
• Remove SDK --json-schema dependence because the gateway/model completes successfully but does not populate Claude Agent SDK structured_output.
• Add scripts/extract_claude_structured_review.py to extract the final review JSON from Claude Code Action's execution_file.
• Keep pull_request_target on trusted default-branch checkout, while allowing workflow_dispatch branch testing to checkout the selected ref.

Why

Manual testing on #198 showed the initial workflow could run but fell back to "AI review unavailable":

• default Claude model caused API Error: 400 Param Incorrect with the gateway;
• after pinning MIMO, Claude completed, but structured_output stayed empty;
• parsing execution_file produced the intended structured AI review result.

Validation

• python3 -m unittest tests.tools.test_extract_claude_structured_review tests.tools.test_pr_ai_review_report tests.tools.test_pr_policy_check tests.tools.test_pr_preflight_summary_comment
• python3 -m py_compile scripts/extract_claude_structured_review.py scripts/pr_ai_review_report.py scripts/pr_policy_check.py scripts/pr_preflight_summary_comment.py
• git diff --check
• ruby -e 'require "yaml"; YAML.load_file(".github/workflows/pr-ai-review.yml"); puts "ok"'
• Manual workflow run on [codex] 修复输入发送后草稿残留 #198 using this branch: https://github.com/memex-lab/memex/actions/runs/26525282624

Result from #198 smoke test

• Risk: low
• Human review required: false
• Golden path impact: possible / record_capture
• Comment and labels were updated on [codex] 修复输入发送后草稿残留 #198 from the real structured review output.

[github-actions]

PR AI Review / PR AI 语义预检

中文

• 风险等级：高风险
• 需要人工审核：是
• 黄金链路影响：可能
• 置信度：high
• Workflow run：26525426633

AI 语义预检未得到可解析结果，需要 maintainer 人工审核。

影响范围

• unknown

黄金链路

• none
• 说明：Claude Code Action 未产出结构化 JSON，AI 语义预检结果不可用。

风险项

• high AI review 结果不可用。证据：Claude Code Action structured_output was empty or invalid.。
  建议：检查 Claude Action 日志和模型网关配置后重新运行 workflow。

测试缺口

• 未发现新的测试缺口。

English

• Risk level: HIGH
• Human review required: YES
• Golden path impact: POSSIBLE
• Confidence: high
• Workflow run: 26525426633

AI semantic review did not return a parseable result; maintainer review is required.

Affected Areas

• unknown

Golden Path

• none
• Rationale: Claude Code Action did not produce structured JSON, so the AI review result is unavailable.

Findings

• high AI review result unavailable. Evidence: Claude Code Action structured_output was empty or invalid..
  Recommendation: Check Claude Action logs and model gateway configuration, then rerun the workflow.

Test Gaps

• No new test gap was reported.

AI review is advisory. Maintainers should verify the result before merging.

[github-actions]

PR Preflight Summary / PR 预检汇总

中文

• 统一结论：高风险：质量预检通过，但需要 maintainer 人工 review 后手动合并。
• Policy preflight：高风险。命中 1 条高风险规则，需要 maintainer 人工确认。
• Flutter quality：通过。Analyzer 和 test baseline 均未发现新增问题。
• PR head：ddeeac1c3bf537b56299fcdcdf4bb72e76052bf2
• Policy run：26525426709
• Flutter run：26525426631

English

• Combined result: High risk: quality passed, but maintainer review is required before manual merge.
• Policy preflight: HIGH RISK. Found 1 high-risk policy signal(s); maintainer review is required.
• Flutter quality: PASS. Analyzer and test baselines found no newly introduced issue.
• PR head: ddeeac1c3bf537b56299fcdcdf4bb72e76052bf2
• Policy run: 26525426709
• Flutter run: 26525426631

PR Policy Preflight / PR 规则预检

PR Policy Preflight / PR 规则预检

中文

• 判定：高风险
• 变更文件数：3
• 变更行数：247
• Diff 是否截断：false

规则命中

• 高风险 github-config .github/workflows/pr-ai-review.yml: GitHub 配置或 workflow 发生变化。

English

• Decision: HIGH RISK
• Changed files: 3
• Changed lines: 247
• Diff truncated: false

Findings

• high github-config .github/workflows/pr-ai-review.yml: GitHub configuration or workflow changed.

PR Flutter Quality / Flutter 质量预检

PR Flutter Quality / Flutter 质量预检

中文

• 总体：通过
• Analyzer baseline：通过
• Test baseline：通过

English

• Overall: PASS
• Analyzer baseline: PASS
• Test baseline: PASS

Flutter Analyzer Baseline

• Base issues: 307
• PR issues: 307
• New issues: 0

No new analyzer issues introduced by this PR.

Flutter Test Baseline

• Base failures: 0
• PR failures: 0
• New failures: 0

No new Flutter test failures introduced by this PR.