[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6ACXZ4D0GXEBX1N35187K8B":3,"skill-same:01M6ACXZ4D0GXEBX1N35187K8B":79},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":39,"linkOnly":40,"markdownContent":41,"markdownBody":42,"frontMatter":43,"refName":44,"commitSha":45,"skillDirectory":46,"forksCount":47,"repositoryHtmlUrl":48,"repositoryHomepageUrl":49,"ownerHtmlUrl":50,"ownerBlogUrl":51,"ownerDescription":6,"updatedAt":52,"install":53,"resources":66},"01M6ACXZ4D0GXEBX1N35187K8B","ansi-analysis",null,"Analyze Gluten ANSI-mode test results (run dev\u002Fverify-ansi-expressions.sh, parse JSON tracker output, produce root-cause analysis and fix recommendations). Trigger on user requests like \"analyze ANSI tests\", \"run ANSI matrix\", \"why is this ANSI test failing\".","apache\u002Fgluten","apache","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F47359?v=4","01M6AC83E6K4FDKK7HZ8WNMQKE","The Apache Software Foundation","01M6ACXWJY7X2YK7EMFJ68JKCA",1623,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fapache\u002Fgluten\u002Fblob\u002Fmain\u002F.github\u002Fskills\u002Fansi-analysis\u002FSKILL.md",[],[19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38],"ansi analysis","analysis","ansi","root-cause","failing","parse","matrix","mode","output","trigger","tests","recommendations","test","tracker","fix","analyze","why","produce","dev","run","2026-11-10 10:12:30",false,"---\nname: ansi-analysis\ndescription: Analyze Gluten ANSI-mode test results (run dev\u002Fverify-ansi-expressions.sh, parse JSON tracker output, produce root-cause analysis and fix recommendations). Trigger on user requests like \"analyze ANSI tests\", \"run ANSI matrix\", \"why is this ANSI test failing\".\n---\n\n# ANSI Test Analysis Skill\n\n## Step 0 — MUST READ FIRST: shared analysis prompt\n\nBefore doing anything else, read the shared prompt that defines the analysis output format and reference source locations:\n\n```\n.github\u002Fskills\u002Fansi-analysis\u002Fshared.md\n```\n\nThis file is the single source of truth — the same content is consumed by the CI Python pipeline (`.github\u002Fskills\u002Fansi-analysis\u002Fanalyze-ansi.py --ai-analysis`). Your output structure, reference source locations, and self-investigation steps MUST follow it. If the file is missing, STOP and tell the user the repo is in a broken state.\n\n## Step 1 — Decide entry point\n\nAsk the user (or infer from request):\n- Run new tests? → Step 2\n- Re-analyze existing JSON in `target\u002Fansi-offload\u002F`? → Step 3\n- Diagnose a single test failure? → Step 4\n\n## Step 2 — Run the verification script\n\n```bash\n.\u002Fdev\u002Fverify-ansi-expressions.sh \u003Ccategory> \u003Cspark41|spark40|all> [--clean]\n```\n\nCategories: `cast | arithmetic | collection | datetime | math | decimal | string | aggregate | errors | all`\n\nLogs: `\u002Ftmp\u002Fansi-matrix\u002Flatest\u002F` (bash logs).\nJSON: `target\u002Fansi-offload\u002F*.json` (written by `GlutenExpressionOffloadTracker.scala`, this is the structured input for analysis).\n\nNotes from prior runs:\n- Use `all` mode in single JVM (~28 min) when full coverage is needed\n- After rebase \u002F branch switch, run `.\u002Fdev\u002Fbuilddep-veloxbe-inc.sh` first to refresh `libvelox.so` \u002F `libgluten.so`\n\n## Step 3 — Analyze JSON results\n\nTwo options:\n\n### 3a. Local AI orchestration (this skill, recommended for interactive review)\n\n1. Read `.github\u002Fskills\u002Fansi-analysis\u002Fshared.md` (Step 0)\n2. List `target\u002Fansi-offload\u002F*.json`\n3. Read each JSON; extract: suite name, total\u002Fpassed\u002Ffailed\u002Fignored counts, per-test `failCause`\n4. Apply the analysis template from shared.md verbatim (sections, tables, constraints)\n5. For each failure: extract Velox file:line from `failCause`, read those C++ files, verify root cause\n6. **Always** grep `isAnsiSupported` in `ep\u002Fbuild-velox\u002Fbuild\u002Fvelox_ep\u002Fvelox\u002Ffunctions\u002Fsparksql\u002Fspecialforms\u002FSparkCastExpr.cpp` when the failure involves Cast — most NO_EXCEPTION\u002FCast failures stem from the small whitelist there\n7. Output the markdown report\n\n### 3b. Python script (CI \u002F batch)\n\n```bash\npython3 .github\u002Fskills\u002Fansi-analysis\u002Fanalyze-ansi.py \\\n  --json-dir target\u002Fansi-offload\u002F \\\n  --ai-analysis \\\n  --output ansi-report.md\n```\n\nThe script loads the same shared prompt and calls the GitHub Models API.\n\n## Step 4 — Single-failure diagnosis\n\nWhen the user pastes one failing test:\n1. Locate its JSON entry under `target\u002Fansi-offload\u002F`\n2. Apply the self-investigation steps from shared.md (extract Velox file:line, check `isAnsiSupported`, cross-check `withAnsiEvalMode` in ExpressionUtils)\n3. Output: Symptom \u002F Root Cause \u002F Fix Point \u002F Representative Tests \u002F Estimated Impact\n\n## Step 5 — Optional PR comment\n\nIf the user wants the report posted to a PR:\n\n```bash\ngh pr comment \u003Cpr-number> --body-file ansi-report.md\n```\n\n(or use the GitHub MCP server tool when available)\n\n## Environment requirements\n\nFor Step 2 (running tests):\n- `SPARK_ANSI_SQL_MODE=true`\n- `SPARK_TESTING=true`\n- `SPARK_SCALA_VERSION=2.13`\n- JVM: `-Dspark.gluten.sql.ansiFallback.enabled=false`\n- Maven profile: include `-Pdelta`\n\n## What NOT to do\n\n- Do NOT invent reference paths or line numbers — always grep \u002F verify\n- Do NOT skip Step 0 — drift between shared.md and your output is the failure mode this skill is designed to prevent\n- Do NOT bypass the shared prompt by writing your own analysis structure\n","# ANSI Test Analysis Skill\n\n## Step 0 — MUST READ FIRST: shared analysis prompt\n\nBefore doing anything else, read the shared prompt that defines the analysis output format and reference source locations:\n\n```\n.github\u002Fskills\u002Fansi-analysis\u002Fshared.md\n```\n\nThis file is the single source of truth — the same content is consumed by the CI Python pipeline (`.github\u002Fskills\u002Fansi-analysis\u002Fanalyze-ansi.py --ai-analysis`). Your output structure, reference source locations, and self-investigation steps MUST follow it. If the file is missing, STOP and tell the user the repo is in a broken state.\n\n## Step 1 — Decide entry point\n\nAsk the user (or infer from request):\n- Run new tests? → Step 2\n- Re-analyze existing JSON in `target\u002Fansi-offload\u002F`? → Step 3\n- Diagnose a single test failure? → Step 4\n\n## Step 2 — Run the verification script\n\n```bash\n.\u002Fdev\u002Fverify-ansi-expressions.sh \u003Ccategory> \u003Cspark41|spark40|all> [--clean]\n```\n\nCategories: `cast | arithmetic | collection | datetime | math | decimal | string | aggregate | errors | all`\n\nLogs: `\u002Ftmp\u002Fansi-matrix\u002Flatest\u002F` (bash logs).\nJSON: `target\u002Fansi-offload\u002F*.json` (written by `GlutenExpressionOffloadTracker.scala`, this is the structured input for analysis).\n\nNotes from prior runs:\n- Use `all` mode in single JVM (~28 min) when full coverage is needed\n- After rebase \u002F branch switch, run `.\u002Fdev\u002Fbuilddep-veloxbe-inc.sh` first to refresh `libvelox.so` \u002F `libgluten.so`\n\n## Step 3 — Analyze JSON results\n\nTwo options:\n\n### 3a. Local AI orchestration (this skill, recommended for interactive review)\n\n1. Read `.github\u002Fskills\u002Fansi-analysis\u002Fshared.md` (Step 0)\n2. List `target\u002Fansi-offload\u002F*.json`\n3. Read each JSON; extract: suite name, total\u002Fpassed\u002Ffailed\u002Fignored counts, per-test `failCause`\n4. Apply the analysis template from shared.md verbatim (sections, tables, constraints)\n5. For each failure: extract Velox file:line from `failCause`, read those C++ files, verify root cause\n6. **Always** grep `isAnsiSupported` in `ep\u002Fbuild-velox\u002Fbuild\u002Fvelox_ep\u002Fvelox\u002Ffunctions\u002Fsparksql\u002Fspecialforms\u002FSparkCastExpr.cpp` when the failure involves Cast — most NO_EXCEPTION\u002FCast failures stem from the small whitelist there\n7. Output the markdown report\n\n### 3b. Python script (CI \u002F batch)\n\n```bash\npython3 .github\u002Fskills\u002Fansi-analysis\u002Fanalyze-ansi.py \\\n  --json-dir target\u002Fansi-offload\u002F \\\n  --ai-analysis \\\n  --output ansi-report.md\n```\n\nThe script loads the same shared prompt and calls the GitHub Models API.\n\n## Step 4 — Single-failure diagnosis\n\nWhen the user pastes one failing test:\n1. Locate its JSON entry under `target\u002Fansi-offload\u002F`\n2. Apply the self-investigation steps from shared.md (extract Velox file:line, check `isAnsiSupported`, cross-check `withAnsiEvalMode` in ExpressionUtils)\n3. Output: Symptom \u002F Root Cause \u002F Fix Point \u002F Representative Tests \u002F Estimated Impact\n\n## Step 5 — Optional PR comment\n\nIf the user wants the report posted to a PR:\n\n```bash\ngh pr comment \u003Cpr-number> --body-file ansi-report.md\n```\n\n(or use the GitHub MCP server tool when available)\n\n## Environment requirements\n\nFor Step 2 (running tests):\n- `SPARK_ANSI_SQL_MODE=true`\n- `SPARK_TESTING=true`\n- `SPARK_SCALA_VERSION=2.13`\n- JVM: `-Dspark.gluten.sql.ansiFallback.enabled=false`\n- Maven profile: include `-Pdelta`\n\n## What NOT to do\n\n- Do NOT invent reference paths or line numbers — always grep \u002F verify\n- Do NOT skip Step 0 — drift between shared.md and your output is the failure mode this skill is designed to prevent\n- Do NOT bypass the shared prompt by writing your own analysis structure\n",{"name":5,"description":7},"main","cce7716962d3152b46fc22882bed4a3887a792ef",".github\u002Fskills\u002Fansi-analysis",679,"https:\u002F\u002Fgithub.com\u002Fapache\u002Fgluten","https:\u002F\u002Fgluten.apache.org\u002F","https:\u002F\u002Fgithub.com\u002Fapache","https:\u002F\u002Fwww.apache.org\u002F","2026-11-14 16:37:36",{"source":8,"skillName":5,"commands":54,"prompt":64,"zipAvailable":65},[55,58,61],{"tool":56,"command":57},"npx","npx skills add apache\u002Fgluten --skill ansi-analysis",{"tool":59,"command":60},"bunx","bunx skills add apache\u002Fgluten --skill ansi-analysis",{"tool":62,"command":63},"pnpm","pnpm dlx skills add apache\u002Fgluten --skill ansi-analysis","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fapache\u002Fgluten\u002Fblob\u002Fmain\u002F.github\u002Fskills\u002Fansi-analysis\u002FSKILL.md",true,[67,73],{"path":68,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":71,"sourceHtmlUrl":72},"analyze-ansi.py","SCRIPT","脚本",22467,"https:\u002F\u002Fgithub.com\u002Fapache\u002Fgluten\u002Fblob\u002Fmain\u002F.github\u002Fskills\u002Fansi-analysis\u002Fanalyze-ansi.py",{"path":74,"resourceType":75,"resourceTypeDesc":76,"sizeBytes":77,"sourceHtmlUrl":78},"shared.md","REFERENCE","参考文档",9976,"https:\u002F\u002Fgithub.com\u002Fapache\u002Fgluten\u002Fblob\u002Fmain\u002F.github\u002Fskills\u002Fansi-analysis\u002Fshared.md",[]]