[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6HWN6PCV3007TBSX6PA18PR":3,"skill-same:01M6HWN6PCV3007TBSX6PA18PR":64},{"skillNo":4,"name":5,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":25,"publishedAt":31,"linkOnly":32,"markdownContent":33,"markdownBody":34,"frontMatter":35,"refName":36,"commitSha":37,"skillDirectory":38,"forksCount":39,"repositoryHtmlUrl":40,"repositoryHomepageUrl":6,"ownerHtmlUrl":41,"ownerBlogUrl":42,"ownerDescription":6,"updatedAt":43,"install":44,"resources":57},"01M6HWN6PCV3007TBSX6PA18PR","pdf-download-extract-fallback-enhanced",null,"HKUDS\u002FAgentPilot","HKUDS","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F118165258?v=4","01M6HW8VRNZP64G0P4MVJW2KKQ","✨Data Intelligence Lab@HKU✨","01M6HW8VWZVHBRHYH3T330PFWV",8029,"MIT","https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fopenspace\u002Fbenchmarks\u002Fgdpval\u002Fskills\u002Fpdf-download-extract-fallback-enhanced\u002FSKILL.md",[17,22],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000002","文档处理","SCENARIO","使用场景",{"categoryNo":23,"categoryName":24,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000023","Skill与智能体开发",[26,27,28,29,30],"PDF","download","extract","fallback","pdf download extract fallback enhanced","2026-11-10 10:12:30",false,"---\nname: pdf-extract-ordered-fallback\ndescription: PDF extraction with ordered tool chain: read_file, then run_shell\u002Fpdftotext, then execute_code_sandbox\u002FPyMuPDF\n---\n\n# PDF Download and Extract with Ordered Fallback\n\nThis skill provides a robust workflow for acquiring PDF documents from web sources and extracting their text content, with a clearly ordered sequence of tool invocations to maximize success rate.\n\n## Overview\n\nWhen working with PDFs from web sources, encounters with JavaScript redirects, corrupted files, missing tools, or inaccessible content are common. This workflow ensures maximum success rate through a严格 ordered fallback sequence that prioritizes shell-based tools over Python sandbox execution.\n\n## Ordered Tool Chain Summary\n\n| Step | Tool | Method | Priority |\n|------|------|--------|----------|\n| 0 | read_file | Direct PDF text extraction | First attempt |\n| 1 | run_shell | pdftotext command | Primary fallback (if Step 0 returns binary\u002Ffails) |\n| 2 | execute_code_sandbox | PyMuPDF Python library | Secondary fallback (if Step 1 fails) |\n| 3 | Domain knowledge | Manual content generation | Last resort |\n\n**Key principle**: Always try shell tools (`run_shell`) before Python sandbox (`execute_code_sandbox`) when both are viable options. Shell execution is more reliable in constrained environments.\n\n## Step-by-Step Instructions\n\n### Step 0: Initial Extraction Attempt with read_file Tool\n\nFirst, attempt to extract PDF text using the `read_file` tool. This is the simplest approach and handles many PDFs correctly:\n\n```\nread_file filetype=\"pdf\" file_path=\"path\u002Fto\u002Fdocument.pdf\"\n```\n\nExpected outcomes:\n- **Success**: Returns extracted text content - proceed to use this directly\n- **Binary\u002FImage data returned**: The tool failed to extract text; file content is raw binary or image data\n  - **Immediate action**: Proceed to Step 1 (run_shell with pdftotext)\n- **Error returned**: Tool failed entirely; proceed to Step 1 (run_shell with pdftotext)\n\n**Critical**: If `read_file` returns binary data (PNG\u002FJPEG headers, raw PDF bytes), do NOT attempt to parse it manually. Immediately switch to shell-based `pdftotext`.\n\n### Step 1: Download PDF with Browser User-Agent\n\nMany PDF hosting sites use JavaScript-based redirects or block automated requests. Use curl with a realistic browser user-agent:\n\n```bash\ncurl -L -A \"Mozilla\u002F5.0 (Windows NT 10.0; Win64; x64) AppleWebKit\u002F537.36 (KHTML, like Gecko) Chrome\u002F120.0.0.0 Safari\u002F537.36\" -o output.pdf \"URL_HERE\"\n```\n\nKey flags:\n- `-L`: Follow redirects\n- `-A`: Set user-agent header to mimic a real browser\n- `-o`: Specify output filename\n\n### Step 2: Verify File Type Before Parsing\n\nAlways validate the downloaded file is actually a PDF before attempting extraction:\n\n```bash\nfile output.pdf\n```\n\nExpected output should contain \"PDF document\". If not:\n- The URL may have redirected to an HTML error page\n- The file may be corrupted\n- Access may be blocked\n\n### Step 3: Primary Shell Extraction with pdftotext via run_shell\n\n**This step takes priority over Python-based extraction.** If Step 0 failed or if you're working with a newly downloaded PDF, use `run_shell` with `pdftotext` before attempting any Python libraries:\n\n```bash\npdftotext downloaded.pdf extracted.txt\n```\n\nExecute via `run_shell`:\n```\nrun_shell command=\"pdftotext downloaded.pdf extracted.txt\"\n```\n\nIf `pdftotext` is not available, install it first:\n```bash\n# Debian\u002FUbuntu\napt-get update && apt-get install -y poppler-utils\n\n# macOS  \nbrew install poppler\n\n# RHEL\u002FCentOS\nyum install -y poppler-utils\n```\n\n**Why shell-first?** Shell-based `pdftotext` is more reliable, faster, and avoids sandbox execution issues that can affect Python code execution in constrained environments.\n\n### Step 4: Secondary Python Fallback with PyMuPDF via execute_code_sandbox\n\nOnly if `run_shell` with `pdftotext` fails or is unavailable, fall back to Python's PyMuPDF library via `execute_code_sandbox`:\n\n```python\nimport fitz  # PyMuPDF\n\ndoc = fitz.open(\"downloaded.pdf\")\ntext = \"\"\nfor page in doc:\n    text += page.get_text()\ndoc.close()\n\nwith open(\"extracted.txt\", \"w\") as f:\n    f.write(text)\n```\n\nExecute within `execute_code_sandbox`:\n```\nexecute_code_sandbox code=\"\u003CPython code above>\"\n```\n\nInstall if needed:\n```bash\npip install pymupdf\n```\n\n**Note**: Some environments may experience `execute_code_sandbox` failures (unknown errors). This is why shell-based extraction (Step 3) must be attempted first.\n\n### Step 5: Graceful Degradation to Domain Knowledge\n\nIf the PDF cannot be accessed or extracted after all attempts:\n\n1. Document the failure mode (network issue, corrupted file, access denied, etc.)\n2. Extract any partial content that was successfully retrieved\n3. Supplement missing content from established domain knowledge\n4. Clearly mark which portions are from source vs. generated from knowledge\n5. Provide citations for any claimed requirements or specifications\n\nExample degradation note:\n```\nNOTE: Source document [URL] was inaccessible due to [reason]. \nContent below combines partial extraction with established domain knowledge \nfor [topic]. Verify against official sources when available.\n```\n\n## Complete Workflow Script\n\n```bash\n#!\u002Fbin\u002Fbash\n# pdf-extract-workflow.sh\n\nPDF_URL=\"$1\"\nOUTPUT_PDF=\"downloaded.pdf\"\nOUTPUT_TXT=\"extracted.txt\"\n\n# Step 0\u002F1: Download with browser user-agent\necho \"Downloading PDF...\"\ncurl -L -A \"Mozilla\u002F5.0 (Windows NT 10.0; Win64; x64) AppleWebKit\u002F537.36\" -o \"$OUTPUT_PDF\" \"$PDF_URL\"\n\n# Step 1: Verify file type\necho \"Verifying file type...\"\nif ! file \"$OUTPUT_PDF\" | grep -q \"PDF document\"; then\n    echo \"WARNING: Downloaded file is not a valid PDF\"\n    echo \"Attempting fallback extraction anyway...\"\nfi\n\n# Step 2: Try pdftotext via shell (PRIMARY EXTRACTION)\necho \"Attempting pdftotext extraction...\"\nif command -v pdftotext &> \u002Fdev\u002Fnull; then\n    if pdftotext \"$OUTPUT_PDF\" \"$OUTPUT_TXT\" 2>\u002Fdev\u002Fnull; then\n        echo \"Extraction successful with pdftotext\"\n        exit 0\n    fi\nfi\n\n# Step 3: Fallback to PyMuPDF via Python sandbox (SECONDARY)\necho \"Falling back to PyMuPDF...\"\npython3 \u003C\u003C 'PYTHON_SCRIPT'\nimport fitz\nimport sys\n\ntry:\n    doc = fitz.open(\"downloaded.pdf\")\n    text = \"\"\n    for page in doc:\n        text += page.get_text()\n    doc.close()\n    with open(\"extracted.txt\", \"w\") as f:\n        f.write(text)\n    print(\"Extraction successful with PyMuPDF\")\n    sys.exit(0)\nexcept Exception as e:\n    print(f\"PyMuPDF failed: {e}\")\n    sys.exit(1)\nPYTHON_SCRIPT\n\n# Step 4: Handle complete failure\nif [ $? -ne 0 ]; then\n    echo \"All extraction methods failed. Generate content from domain knowledge.\"\n    echo \"Document the failure and proceed with knowledge-based content generation.\"\nfi\n```\n\n## Agent-Specific Tool Invocation Pattern\n\nFor AI agents with access to specialized tools, follow this exact sequence:\n\n```\n# ITERATION 1: Try read_file first\nread_file filetype=\"pdf\" file_path=\"document.pdf\"\n\n# If read_file returns binary data or fails:\n# ITERATION 2: Use run_shell with pdftotext\nrun_shell command=\"pdftotext document.pdf extracted.txt\"\n\n# If run_shell fails:\n# ITERATION 3: Use execute_code_sandbox with PyMuPDF\nexecute_code_sandbox code=\"import fitz; doc = fitz.open('document.pdf'); ...\"\n\n# If all automated extraction fails:\n# ITERATION 4+: Document failure mode and generate from domain knowledge\n```\n\n**Critical anti-pattern to avoid**: Do NOT attempt `execute_code_sandbox` before `run_shell` for PDF extraction. Shell tools are more reliable and should be prioritized.\n\n## Failure Recovery from Observed Patterns\n\nBased on execution analysis, here are common failure cascades and recovery strategies:\n\n### Pattern: read_file Returns Binary Data\n**Symptom**: `read_file` returns PNG\u002FJPEG image data or raw PDF bytes instead of text\n**Cause**: Tool cannot extract text from scanned PDFs or certain PDF structures  \n**Recovery**: Immediately switch to `run_shell` with `pdftotext` - do not attempt to parse binary data\n\n### Pattern: execute_code_sandbox Returns Unknown Error\n**Symptom**: Multiple `execute_code_sandbox` calls fail with \"unknown error\"\n**Cause**: Sandbox execution environment issues or resource constraints\n**Recovery**: This is why `run_shell` must be attempted first - shell execution bypasses sandbox limitations\n\n### Pattern: read_webpage Fails on All URLs\n**Symptom**: All `read_webpage` calls to domain URLs return errors\n**Cause**: Anti-bot measures, network issues, or site blocking\n**Recovery**: Focus on PDF extraction from locally downloaded files; supplement missing context from domain knowledge with clear citations\n\n## Best Practices\n\n1. **Always verify before parsing**: Never assume a downloaded file is valid\n2. **Preserve original PDF**: Keep the downloaded file for debugging if needed\n3. **Log each step**: Document which method succeeded for future reference\n4. **Check extraction quality**: Verify extracted text is readable and complete\n5. **Cite source limitations**: When using fallback knowledge, clearly indicate source gaps\n6. **Follow tool ordering**: read_file → run_shell → execute_code_sandbox → domain knowledge\n7. **Shell before Python**: Prioritize `run_shell` over `execute_code_sandbox` when both are viable\n8. **Detect binary early**: If `read_file` returns non-text data, immediately switch to shell tools\n\n## Common Failure Modes\n\n| Symptom | Cause | Solution |\n|---------|-------|----------|\n| HTML content in PDF | URL redirected to error page | Check HTTP status, try alternate URL |\n| Empty extraction | Password-protected or scanned PDF | Try OCR tools or request accessible version |\n| Garbled text | Encoding issues | Try PyMuPDF with different extraction mode |\n| read_file returns binary | Scanned PDF or tool limitation | Immediately use run_shell with pdftotext |\n| execute_code_sandbox unknown error | Sandbox execution failure | This is why run_shell should be tried first |\n| Curl blocked | Anti-bot measures | Add more headers, use delay between requests |\n\n## When to Use This Skill\n\n- Downloading regulatory documents from government websites\n- Extracting content from technical manuals or handbooks\n- Processing PDFs in automated pipelines where reliability matters\n- Situations where tool execution constraints may limit Python sandbox availability\n- Any situation where PDF access may be unreliable or restricted\n","# PDF Download and Extract with Ordered Fallback\n\nThis skill provides a robust workflow for acquiring PDF documents from web sources and extracting their text content, with a clearly ordered sequence of tool invocations to maximize success rate.\n\n## Overview\n\nWhen working with PDFs from web sources, encounters with JavaScript redirects, corrupted files, missing tools, or inaccessible content are common. This workflow ensures maximum success rate through a严格 ordered fallback sequence that prioritizes shell-based tools over Python sandbox execution.\n\n## Ordered Tool Chain Summary\n\n| Step | Tool | Method | Priority |\n|------|------|--------|----------|\n| 0 | read_file | Direct PDF text extraction | First attempt |\n| 1 | run_shell | pdftotext command | Primary fallback (if Step 0 returns binary\u002Ffails) |\n| 2 | execute_code_sandbox | PyMuPDF Python library | Secondary fallback (if Step 1 fails) |\n| 3 | Domain knowledge | Manual content generation | Last resort |\n\n**Key principle**: Always try shell tools (`run_shell`) before Python sandbox (`execute_code_sandbox`) when both are viable options. Shell execution is more reliable in constrained environments.\n\n## Step-by-Step Instructions\n\n### Step 0: Initial Extraction Attempt with read_file Tool\n\nFirst, attempt to extract PDF text using the `read_file` tool. This is the simplest approach and handles many PDFs correctly:\n\n```\nread_file filetype=\"pdf\" file_path=\"path\u002Fto\u002Fdocument.pdf\"\n```\n\nExpected outcomes:\n- **Success**: Returns extracted text content - proceed to use this directly\n- **Binary\u002FImage data returned**: The tool failed to extract text; file content is raw binary or image data\n  - **Immediate action**: Proceed to Step 1 (run_shell with pdftotext)\n- **Error returned**: Tool failed entirely; proceed to Step 1 (run_shell with pdftotext)\n\n**Critical**: If `read_file` returns binary data (PNG\u002FJPEG headers, raw PDF bytes), do NOT attempt to parse it manually. Immediately switch to shell-based `pdftotext`.\n\n### Step 1: Download PDF with Browser User-Agent\n\nMany PDF hosting sites use JavaScript-based redirects or block automated requests. Use curl with a realistic browser user-agent:\n\n```bash\ncurl -L -A \"Mozilla\u002F5.0 (Windows NT 10.0; Win64; x64) AppleWebKit\u002F537.36 (KHTML, like Gecko) Chrome\u002F120.0.0.0 Safari\u002F537.36\" -o output.pdf \"URL_HERE\"\n```\n\nKey flags:\n- `-L`: Follow redirects\n- `-A`: Set user-agent header to mimic a real browser\n- `-o`: Specify output filename\n\n### Step 2: Verify File Type Before Parsing\n\nAlways validate the downloaded file is actually a PDF before attempting extraction:\n\n```bash\nfile output.pdf\n```\n\nExpected output should contain \"PDF document\". If not:\n- The URL may have redirected to an HTML error page\n- The file may be corrupted\n- Access may be blocked\n\n### Step 3: Primary Shell Extraction with pdftotext via run_shell\n\n**This step takes priority over Python-based extraction.** If Step 0 failed or if you're working with a newly downloaded PDF, use `run_shell` with `pdftotext` before attempting any Python libraries:\n\n```bash\npdftotext downloaded.pdf extracted.txt\n```\n\nExecute via `run_shell`:\n```\nrun_shell command=\"pdftotext downloaded.pdf extracted.txt\"\n```\n\nIf `pdftotext` is not available, install it first:\n```bash\n# Debian\u002FUbuntu\napt-get update && apt-get install -y poppler-utils\n\n# macOS  \nbrew install poppler\n\n# RHEL\u002FCentOS\nyum install -y poppler-utils\n```\n\n**Why shell-first?** Shell-based `pdftotext` is more reliable, faster, and avoids sandbox execution issues that can affect Python code execution in constrained environments.\n\n### Step 4: Secondary Python Fallback with PyMuPDF via execute_code_sandbox\n\nOnly if `run_shell` with `pdftotext` fails or is unavailable, fall back to Python's PyMuPDF library via `execute_code_sandbox`:\n\n```python\nimport fitz  # PyMuPDF\n\ndoc = fitz.open(\"downloaded.pdf\")\ntext = \"\"\nfor page in doc:\n    text += page.get_text()\ndoc.close()\n\nwith open(\"extracted.txt\", \"w\") as f:\n    f.write(text)\n```\n\nExecute within `execute_code_sandbox`:\n```\nexecute_code_sandbox code=\"\u003CPython code above>\"\n```\n\nInstall if needed:\n```bash\npip install pymupdf\n```\n\n**Note**: Some environments may experience `execute_code_sandbox` failures (unknown errors). This is why shell-based extraction (Step 3) must be attempted first.\n\n### Step 5: Graceful Degradation to Domain Knowledge\n\nIf the PDF cannot be accessed or extracted after all attempts:\n\n1. Document the failure mode (network issue, corrupted file, access denied, etc.)\n2. Extract any partial content that was successfully retrieved\n3. Supplement missing content from established domain knowledge\n4. Clearly mark which portions are from source vs. generated from knowledge\n5. Provide citations for any claimed requirements or specifications\n\nExample degradation note:\n```\nNOTE: Source document [URL] was inaccessible due to [reason]. \nContent below combines partial extraction with established domain knowledge \nfor [topic]. Verify against official sources when available.\n```\n\n## Complete Workflow Script\n\n```bash\n#!\u002Fbin\u002Fbash\n# pdf-extract-workflow.sh\n\nPDF_URL=\"$1\"\nOUTPUT_PDF=\"downloaded.pdf\"\nOUTPUT_TXT=\"extracted.txt\"\n\n# Step 0\u002F1: Download with browser user-agent\necho \"Downloading PDF...\"\ncurl -L -A \"Mozilla\u002F5.0 (Windows NT 10.0; Win64; x64) AppleWebKit\u002F537.36\" -o \"$OUTPUT_PDF\" \"$PDF_URL\"\n\n# Step 1: Verify file type\necho \"Verifying file type...\"\nif ! file \"$OUTPUT_PDF\" | grep -q \"PDF document\"; then\n    echo \"WARNING: Downloaded file is not a valid PDF\"\n    echo \"Attempting fallback extraction anyway...\"\nfi\n\n# Step 2: Try pdftotext via shell (PRIMARY EXTRACTION)\necho \"Attempting pdftotext extraction...\"\nif command -v pdftotext &> \u002Fdev\u002Fnull; then\n    if pdftotext \"$OUTPUT_PDF\" \"$OUTPUT_TXT\" 2>\u002Fdev\u002Fnull; then\n        echo \"Extraction successful with pdftotext\"\n        exit 0\n    fi\nfi\n\n# Step 3: Fallback to PyMuPDF via Python sandbox (SECONDARY)\necho \"Falling back to PyMuPDF...\"\npython3 \u003C\u003C 'PYTHON_SCRIPT'\nimport fitz\nimport sys\n\ntry:\n    doc = fitz.open(\"downloaded.pdf\")\n    text = \"\"\n    for page in doc:\n        text += page.get_text()\n    doc.close()\n    with open(\"extracted.txt\", \"w\") as f:\n        f.write(text)\n    print(\"Extraction successful with PyMuPDF\")\n    sys.exit(0)\nexcept Exception as e:\n    print(f\"PyMuPDF failed: {e}\")\n    sys.exit(1)\nPYTHON_SCRIPT\n\n# Step 4: Handle complete failure\nif [ $? -ne 0 ]; then\n    echo \"All extraction methods failed. Generate content from domain knowledge.\"\n    echo \"Document the failure and proceed with knowledge-based content generation.\"\nfi\n```\n\n## Agent-Specific Tool Invocation Pattern\n\nFor AI agents with access to specialized tools, follow this exact sequence:\n\n```\n# ITERATION 1: Try read_file first\nread_file filetype=\"pdf\" file_path=\"document.pdf\"\n\n# If read_file returns binary data or fails:\n# ITERATION 2: Use run_shell with pdftotext\nrun_shell command=\"pdftotext document.pdf extracted.txt\"\n\n# If run_shell fails:\n# ITERATION 3: Use execute_code_sandbox with PyMuPDF\nexecute_code_sandbox code=\"import fitz; doc = fitz.open('document.pdf'); ...\"\n\n# If all automated extraction fails:\n# ITERATION 4+: Document failure mode and generate from domain knowledge\n```\n\n**Critical anti-pattern to avoid**: Do NOT attempt `execute_code_sandbox` before `run_shell` for PDF extraction. Shell tools are more reliable and should be prioritized.\n\n## Failure Recovery from Observed Patterns\n\nBased on execution analysis, here are common failure cascades and recovery strategies:\n\n### Pattern: read_file Returns Binary Data\n**Symptom**: `read_file` returns PNG\u002FJPEG image data or raw PDF bytes instead of text\n**Cause**: Tool cannot extract text from scanned PDFs or certain PDF structures  \n**Recovery**: Immediately switch to `run_shell` with `pdftotext` - do not attempt to parse binary data\n\n### Pattern: execute_code_sandbox Returns Unknown Error\n**Symptom**: Multiple `execute_code_sandbox` calls fail with \"unknown error\"\n**Cause**: Sandbox execution environment issues or resource constraints\n**Recovery**: This is why `run_shell` must be attempted first - shell execution bypasses sandbox limitations\n\n### Pattern: read_webpage Fails on All URLs\n**Symptom**: All `read_webpage` calls to domain URLs return errors\n**Cause**: Anti-bot measures, network issues, or site blocking\n**Recovery**: Focus on PDF extraction from locally downloaded files; supplement missing context from domain knowledge with clear citations\n\n## Best Practices\n\n1. **Always verify before parsing**: Never assume a downloaded file is valid\n2. **Preserve original PDF**: Keep the downloaded file for debugging if needed\n3. **Log each step**: Document which method succeeded for future reference\n4. **Check extraction quality**: Verify extracted text is readable and complete\n5. **Cite source limitations**: When using fallback knowledge, clearly indicate source gaps\n6. **Follow tool ordering**: read_file → run_shell → execute_code_sandbox → domain knowledge\n7. **Shell before Python**: Prioritize `run_shell` over `execute_code_sandbox` when both are viable\n8. **Detect binary early**: If `read_file` returns non-text data, immediately switch to shell tools\n\n## Common Failure Modes\n\n| Symptom | Cause | Solution |\n|---------|-------|----------|\n| HTML content in PDF | URL redirected to error page | Check HTTP status, try alternate URL |\n| Empty extraction | Password-protected or scanned PDF | Try OCR tools or request accessible version |\n| Garbled text | Encoding issues | Try PyMuPDF with different extraction mode |\n| read_file returns binary | Scanned PDF or tool limitation | Immediately use run_shell with pdftotext |\n| execute_code_sandbox unknown error | Sandbox execution failure | This is why run_shell should be tried first |\n| Curl blocked | Anti-bot measures | Add more headers, use delay between requests |\n\n## When to Use This Skill\n\n- Downloading regulatory documents from government websites\n- Extracting content from technical manuals or handbooks\n- Processing PDFs in automated pipelines where reliability matters\n- Situations where tool execution constraints may limit Python sandbox availability\n- Any situation where PDF access may be unreliable or restricted\n",{},"main","6b84b0689ca0d9e7b1e4061188ff6b60758e0f8a","openspace\u002Fbenchmarks\u002Fgdpval\u002Fskills\u002Fpdf-download-extract-fallback-enhanced",944,"https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot","https:\u002F\u002Fgithub.com\u002FHKUDS","https:\u002F\u002Fsites.google.com\u002Fview\u002Fchaoh","2026-11-12 22:26:39",{"source":7,"skillName":5,"commands":45,"prompt":55,"zipAvailable":56},[46,49,52],{"tool":47,"command":48},"npx","npx skills add HKUDS\u002FAgentPilot --skill pdf-download-extract-fallback-enhanced",{"tool":50,"command":51},"bunx","bunx skills add HKUDS\u002FAgentPilot --skill pdf-download-extract-fallback-enhanced",{"tool":53,"command":54},"pnpm","pnpm dlx skills add HKUDS\u002FAgentPilot --skill pdf-download-extract-fallback-enhanced","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fopenspace\u002Fbenchmarks\u002Fgdpval\u002Fskills\u002Fpdf-download-extract-fallback-enhanced\u002FSKILL.md",true,[58],{"path":59,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":62,"sourceHtmlUrl":63},".skill_id","OTHER","其他",52,"https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fopenspace\u002Fbenchmarks\u002Fgdpval\u002Fskills\u002Fpdf-download-extract-fallback-enhanced\u002F.skill_id",[65,85,104,125,149,172],{"skillNo":66,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":31},"01M6HYB297NDABRERJRVG0K8R0","agentpilot","Set up AgentPilot and use its single entry to prepare missing inputs and run configured RSI experiments. Use when the user requests AgentPilot installation or work with AgentPilot.","https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fplugins\u002Fagentpilot\u002Fskills\u002Fagentpilot\u002FSKILL.md",[],[67,72,73,74,75,76,77,78,79,80,81,82,83,84],"prepare","requests","with","missing","installation","rsi","up","use","user","work","SET","run","experiments",{"skillNo":86,"name":87,"nameZh":6,"description":88,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":31},"01M6HYAYWXTTE1J7DJM0MFNQ4H","remember","Review auto-memory entries and propose promotions to OPENSPACE.md, OPENSPACE.local.md, or retained auto-memory topics; detect duplicates, stale facts, conflicts, and ambiguous entries before making changes.","https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fopenspace\u002Fopenspace\u002Fskills\u002Fremember\u002FSKILL.md",[],[87,92,93,94,95,96,97,98,99,100,101,102,103],"memory","review","auto","openspace","changes","facts","detect","entries","local","conflicts","stale","duplicates",{"skillNo":105,"name":106,"nameZh":6,"description":107,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":108,"categories":109,"keywords":111,"publishedAt":31},"01M6HYAVMPM4SBDV60R2B5HPV8","skill-discovery","Search reusable OpenSpace skills locally, and browse cloud packages step by step when a cloud skill may help. Reusing proven skills saves tokens, improves reliability, and extends your capabilities beyond built-in tools.","https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fopenspace\u002Fopenspace\u002Fhost_skills\u002Fskill-discovery\u002FSKILL.md",[110],{"categoryNo":23,"categoryName":24,"taxonomyType":20,"taxonomyTypeDesc":21},[112,113,114,95,115,116,117,118,119,120,121,122,123,124],"skill discovery","discovery","reusable","step by step","tokens","packages","help","cloud","reliability","capabilities","browse","search","Tools",{"skillNo":126,"name":127,"nameZh":6,"description":128,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":129,"categories":130,"keywords":132,"publishedAt":31},"01M6HYAR811B3RB0M50RBEDGN6","delegate-task","Delegate tasks to OpenSpace — a full-stack autonomous worker for coding, DevOps, web research, and desktop automation, backed by an extensive MCP tool and skill library. Skills auto-improve through use, reducing token consumption over time. A cloud community lets agents share and collectively evolve reusable skills.","https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fopenspace\u002Fopenspace\u002Fhost_skills\u002Fdelegate-task\u002FSKILL.md",[131],{"categoryNo":23,"categoryName":24,"taxonomyType":20,"taxonomyTypeDesc":21},[133,134,135,95,136,137,138,139,140,114,94,79,141,142,143,144,145,146,147,148],"delegate","task","delegate task","time","improve","desktop automation","worker","community","autonomous","tool","stack","for","share","tasks","automation","MCP",{"skillNo":150,"name":151,"nameZh":6,"description":152,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":153,"categories":154,"keywords":155,"publishedAt":31},"01M6HYAM9FSGV6E52AYENDQXYH","write-special-chars-ts-check","Workaround for write_file failures caused by special characters (apostrophes, backticks, template literals) using a Python heredoc script, plus a reliable TypeScript type-checking step that filters output to only the newly-written files to avoid being misled by pre-existing errors.","https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fopenspace\u002Flegacy\u002Fv1\u002Fshowcase\u002Fskills\u002Fwrite-special-chars-ts-check\u002FSKILL.md",[],[156,157,158,159,160,161,162,163,164,165,166,167,168,169,144,170,171],"write special chars ts check","write","check","type","files","filters","only","template","file","python","typescript","output","Existing","type-checking","errors","script",{"skillNo":173,"name":174,"nameZh":6,"description":175,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":176,"categories":177,"keywords":178,"publishedAt":31},"01M6HYAH05HGV7SVWSVR8PJRQD","write-special-chars-readback","Workaround for write_file failures caused by special characters (apostrophes, backticks, template literals) using a Python heredoc script, with mandatory post-write verification via read_file and cleanup of \u002Ftmp helper scripts.","https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fopenspace\u002Flegacy\u002Fv1\u002Fshowcase\u002Fskills\u002Fwrite-special-chars-readback\u002FSKILL.md",[],[179,180,157,74,144,171,181,182,183,184,185,163,164,186,187,165],"write special chars readback","readback","helper","post","via","scripts","read","verification","cleanup"]