USPTO Final Petition Decisions MCP Server

""" Tool reflections and comprehensive guidance for USPTO Final Petition Decisions MCP. This module contains the detailed guidance, workflows, and cross-MCP integration patterns that are returned by the FPD_get_guidance tool (sectioned approach). """ def get_guidance_section(section: str = "overview") -> str: """ Get selective USPTO FPD guidance sections for context-efficient workflows. Args: section: Section name (default: "overview") Returns: Markdown-formatted string for requested section """ sections = { "overview": _get_overview_section(), "workflows_pfw": _get_workflows_pfw_section(), "workflows_ptab": _get_workflows_ptab_section(), "workflows_citations": _get_workflows_citations_section(), "workflows_complete": _get_workflows_complete_section(), "workflows_assistant": _get_workflows_assistant_section(), "tools": _get_tools_section(), "red_flags": _get_red_flags_section(), "documents": _get_documents_section(), "ultra_context": _get_ultra_context_section(), "cost": _get_cost_section() } if section not in sections: return f"Error: Section '{section}' not found. Available sections: {', '.join(sections.keys())}" return sections[section] def _get_overview_section() -> str: """Overview section with quick reference chart and section guide""" return """# USPTO Final Petition Decisions MCP - Tool Guidance **Version:** 3.0 **Last Updated:** 2025-11-02 ## Available Sections and Quick Reference ### 🎯 Quick Reference Chart - What section for your question? - 🔍 **"Find petitions by company/art unit"** → `tools` - 🚩 **"Identify petition red flags"** → `red_flags` - 📄 **"Download petition documents"** → `documents` - 🤝 **"Correlate petitions with prosecution"** → `workflows_pfw` - ⚖️ **"Analyze petition + PTAB patterns"** → `workflows_ptab` - 📊 **"Citation quality + petition correlation"** → `workflows_citations` - 🏢 **"Complete portfolio due diligence"** → `workflows_complete` - 📚 **"Research CFR rules with Assistant"** → `workflows_assistant` - 🎯 **"Ultra-minimal PFW + FPD workflows"** → `ultra_context` - 💰 **"Reduce extraction costs"** → `cost` ### Available Sections: - **overview**: Available sections and MCP overview (this section) - **workflows_pfw**: FPD + PFW integration workflows - **workflows_ptab**: FPD + PTAB integration workflows - **workflows_citations**: FPD + Citations integration workflows - **workflows_complete**: Four-MCP complete lifecycle analysis - **workflows_assistant**: Pinecone Assistant + FPD research workflows - **tools**: Tool catalog, progressive disclosure, parameters - **red_flags**: Petition red flag indicators and CFR rules - **documents**: Document extraction, downloads, proxy configuration - **ultra_context**: PFW fields parameter + ultra-minimal workflows - **cost**: Cost optimization for document extraction ### Context Efficiency Benefits: - **80-95% token reduction** (2-8KB per section vs 62KB total) - **Targeted guidance** for specific workflows - **Same comprehensive content** organized for efficiency - **Consistent pattern** with PFW MCP ## MCP Overview **Name:** USPTO Final Petition Decisions MCP **Purpose:** Track prosecution-related petitions (37 CFR 1.181, 1.182, 1.183) **Position in Lifecycle:** Bridges normal prosecution (PFW) and post-grant challenges (PTAB) **Data Source:** USPTO Open Data Portal - Final Petition Decisions API **Authentication:** Same USPTO_API_KEY as Patent File Wrapper MCP ## Available Tools Summary ### Search Tools - **fpd_search_petitions_minimal**: Ultra-fast discovery (50-100 petitions, 8 fields, 95-99% reduction) - **fpd_search_petitions_balanced**: Detailed analysis (10-20 petitions, 18 fields, 80-88% reduction) - **fpd_search_petitions_by_art_unit**: Art unit quality assessment - **fpd_search_petitions_by_application**: Complete petition history for application ### Detail Tools - **fpd_get_petition_details**: Full petition details with optional documents - **fpd_get_document_download**: Browser-accessible PDF downloads via proxy - **fpd_get_document_content**: Text extraction from petition PDFs (hybrid PyPDF2 + Mistral OCR) ### Guidance Tool - **FPD_get_guidance**: Context-efficient sectioned guidance (this tool) """ def _get_workflows_pfw_section() -> str: """FPD + PFW integration workflows""" return """## FPD + PFW Integration Workflows ### Workflow 1: Complete Patent Lifecycle Tracking **Scenario:** Due diligence on target company portfolio **Steps:** 1. **PFW:** Get company's patent portfolio ```python pfw_search_applications_minimal(applicant_name='Target Co', limit=100) ``` 2. **FPD:** Check petition history for procedural issues ```python fpd_search_petitions_by_application(application_number='17896175') ``` 3. Identify patterns: Abandonments, examiner disputes, procedural problems 4. **PTAB:** Cross-check PTAB challenge history for granted patents 5. Analyze: Prosecution quality with petition context **Value:** Holistic view of patent prosecution and challenge history ### Workflow 2: Abandonment and Revival Analysis **Scenario:** Assessing patents with revival history **Steps:** 1. **FPD:** Find revival petitions (37 CFR 1.137) ```python fpd_search_petitions_balanced(petition_type_code='551', limit=50) ``` 2. **PFW:** Get full prosecution history to understand abandonment reason ```python pfw_search_applications_minimal(application_number='17896175') ``` 3. Categorize: Missed deadline vs. strategic abandonment 4. **PFW:** Check post-revival prosecution quality 5. Context: Understand procedural issues in portfolio **Value:** Understand IP management quality and procedural compliance ### Workflow 3: Art Unit Procedural Analysis **Scenario:** Identifying art units with petition patterns **Steps:** 1. **FPD:** Get all petitions for art unit ```python fpd_search_petitions_by_art_unit(art_unit='2128', date_range='2020-01-01:2024-12-31') ``` 2. Categorize: Examiner disputes, restriction appeals, rule waivers 3. **PFW:** Cross-reference to get examiner names ```python pfw_search_applications_minimal(art_unit='2128', fields=['applicationNumberText', 'examinerNameText'], limit=100) ``` 4. Analyze: Petition frequency and types by examiner **Value:** Understand procedural patterns in prosecution ### Learning PFW Workflows **Context-Efficient Guidance Tool:** `pfw_get_guidance` When working with cross-MCP workflows, use PFW's selective guidance tool: **Relevant Sections for FPD Users:** - `workflows_fpd`: FPD+PFW integration workflows, petition red flag analysis - `workflows_complete`: Complete company due diligence across PFW, FPD, PTAB - `tools`: PFW convenience parameter searches - `fields`: Ultra-minimal field selection for 99% context reduction - `cost`: API cost optimization strategies **Quick Reference:** ```python # Learn FPD+PFW workflows pfw_get_guidance('workflows_fpd') # Learn complete due diligence workflows pfw_get_guidance('workflows_complete') ``` ### PFW XML Data Retrieval - Token Optimization **IMPORTANT:** If you need detailed patent/application XML data from PFW (beyond search metadata), use the `pfw_get_patent_or_application_xml` tool with `include_raw_xml=False` for massive token savings. **Token Reduction:** - With `include_raw_xml=True` (default): ~55KB per application (~50K chars raw XML) - With `include_raw_xml=False`: ~5KB per application (91% reduction) - With `include_raw_xml=False` + `include_fields`: ~500 chars (99% reduction) **Best Practice:** ```python # RECOMMENDED: Exclude raw XML (91% token reduction) pfw_get_patent_or_application_xml( application_number='17896175', include_raw_xml=False # ← CRITICAL for cross-MCP workflows ) # EVEN BETTER: Selective fields (99% reduction) pfw_get_patent_or_application_xml( application_number='17896175', include_raw_xml=False, include_fields=['title', 'abstract', 'claims', 'assignee'] ) ``` **Why This Matters for FPD Workflows:** - Cross-MCP workflows share context budget across multiple MCPs - Raw XML contains ~50K chars of unparsed data most workflows don't need - Using `include_raw_xml=False` allows 10x more applications in same context window - Perfect for: Due diligence, portfolio analysis, petition correlation studies **When to Use Each Approach:** - **Search tools only** (`pfw_search_applications_minimal`): Entity lookup, portfolio discovery (current FPD workflows) - **XML with `include_raw_xml=False`**: Detailed prosecution analysis, claim review - **XML with `include_fields`**: Ultra-targeted data extraction, large-scale analysis """ def _get_workflows_ptab_section() -> str: """FPD + PTAB integration workflows""" return """## FPD + PTAB Integration Workflows ### Three-MCP Availability All three USPTO MCPs support cross-referencing: 1. **PFW (Patent File Wrapper)** - Filing and prosecution history 2. **FPD (Final Petition Decisions)** - Petition context and procedural issues 3. **PTAB (Patent Trial and Appeal Board)** - Post-grant challenges **Complete Lifecycle:** PFW (filing) → FPD (petitions) → Grant → PTAB (challenges) **Integration Value:** Each MCP provides different aspects of patent lifecycle ### Workflow: Prosecution Quality → PTAB Vulnerability Correlation **Scenario:** Assess if petition patterns predict PTAB challenge vulnerability **Steps:** 1. **PFW:** Get granted patents for target company ```python pfw_search_applications_minimal(applicant_name='Target', status='Patented', limit=100) ``` 2. **FPD:** Check petition history during prosecution ```python fpd_search_petitions_by_application(application_number='17896175') ``` 3. **PTAB:** Check for post-grant challenges ```python ptab_search_proceedings_minimal(patent_number='11788453') ``` 4. Correlation Analysis: - Patents with denied examiner dispute petitions → Higher PTAB vulnerability - Patents with revival petitions → Procedural risk indicators - Multiple petitions during prosecution → Quality concerns **Red Flags:** - Denied 37 CFR 1.181 petitions (examiner disputes) + PTAB challenge = Examination quality issues - Revival petitions + PTAB IPR = Procedural and substantive weakness - High petition frequency + PTAB institution = Pattern of prosecution problems **Value:** Predict PTAB vulnerability based on prosecution petition patterns """ def _get_workflows_citations_section() -> str: """FPD + Citations integration workflows""" return """## FPD + Citations Integration Workflows ### Overview Enhanced Citations MCP provides AI-extracted citation intelligence from USPTO Office Actions (Oct 1, 2017 - present). **Data Coverage:** Office Actions MAILED from October 1, 2017 to 30 days prior **Effective Filing Date Coverage:** Applications filed 2015-2016+ typically have citation data **Context Reduction:** 90-95% through progressive disclosure ### Workflow 1: Art Unit Petition Quality with Citation Intelligence **Scenario:** Art unit 2128 has high petition rate - assess if citation quality is a factor **Steps:** 1. **PFW Discovery:** Get art unit applications ```python pfw_search_applications_minimal(art_unit='2128', filing_date_start='2015-01-01', fields=['applicationNumberText', 'examinerNameText'], limit=100) ``` 2. **FPD Petition Patterns:** Get art unit petitions ```python fpd_search_petitions_by_art_unit(art_unit='2128') ``` 3. Calculate baseline petition rate 4. **Citation Analysis:** For 20-30 applications ```python search_citations_minimal(application_number=app_num, rows=50) ``` 5. Aggregate citation metrics: - Citation density (citations per application) - Examiner citation ratio (examinerCitedReferenceIndicator=true / total) - Citation category distribution 6. Correlation Analysis: - LOW citation quality + HIGH petition rate = Art unit quality issues - Normal citations + high petitions = Complex technology **Red Flags:** - Examiner citation ratio below 50% (inadequate search) - Low citation density (<5 citations) in citation-heavy tech - High petition rate (>15%) correlating with low citation quality ### Workflow 2: Examiner Dispute Petitions with Citation Correlation **Scenario:** 37 CFR 1.181 petitions - check if citation patterns indicate examiner issues **Critical:** MUST start with PFW (examiner name not in FPD or Citations APIs) **Steps:** 1. **PFW Examiner Discovery:** ```python pfw_search_applications_minimal(examiner_name='Smith, John', filing_date_start='2015-01-01', fields=['applicationNumberText'], limit=100) ``` 2. **FPD Dispute Identification:** Check each application for petitions 3. Categorize: DISPUTED (has 1.181 petition) vs NON-DISPUTED 4. **Citation Analysis:** Compare citation metrics between groups 5. Correlation patterns: - DISPUTED apps with LOWER examiner citation rate → Inadequate search - DISPUTED apps normal citations → Disputes unrelated to search quality **Citation Quality Indicators:** - Healthy: Examiner citation ratio above 60%, 10+ citations per app - Concern: Examiner citation ratio below 50% - Critical: Below 30% + high dispute rate + granted petitions ### When to Use Citations Integration **Use citations when:** - Analyzing art unit quality with comprehensive assessment - Investigating examiner dispute petitions for root cause - Applications filed 2015-2016+ (likely have Office Action data) **Skip citations when:** - Applications filed before 2015 (unlikely to have data) - Only need petition red flags without citation context - Time/token budget limited """ def _get_workflows_complete_section() -> str: """Complete four-MCP lifecycle workflows""" return """## Complete Four-MCP Lifecycle Analysis ### Complete M&A Due Diligence **Scenario:** Comprehensive patent intelligence across all USPTO databases **Four-MCP Integration Workflow:** 1. **Portfolio Discovery (PFW)** ```python pfw_search_applications_minimal(applicant_name='Target Company', filing_date_start='2015-01-01', limit=100) ``` 2. **Citation Intelligence (Citations)** - For applications filed 2015-2016+ ```python search_citations_minimal(application_number=app_num, rows=50) ``` Analyze examiner citation patterns for prosecution quality 3. **FPD Risk Assessment (FPD)** ```python fpd_search_petitions_by_application(application_number=app_num) ``` Check procedural irregularities and petition history 4. **PTAB Challenge Analysis (PTAB)** - For granted patents ```python ptab_search_proceedings_minimal(patent_number=patent_num) ``` Assess post-grant challenge exposure 5. **Document Intelligence (PFW)** Extract key prosecution documents for detailed analysis 6. **Comprehensive Reporting** Integrate findings across all four data sources **Enhanced Risk Scoring Matrix:** - **Technical Strength**: Claim scope, prosecution quality, prior art landscape - **Legal Enforceability**: Citation thoroughness, procedural cleanliness - **Challenge Exposure**: PTAB proceedings history and outcomes - **Procedural Issues**: FPD petition patterns and denial history ### Patent Lifecycle Stages and MCP Coverage **Stage 1: Filing (PFW)** - Application filing and initial prosecution - Examiner assignments and office actions - Amendments and responses **Stage 2: Procedural Issues (FPD)** - Revival petitions (missed deadlines) - Examiner disputes (supervisory review) - Restriction challenges **Stage 3: Citation Intelligence (Citations)** - Oct 2017+ - Examiner search quality assessment - Prior art thoroughness - Citation category analysis **Stage 4: Grant (PFW)** - Patent issuance - Final prosecution history **Stage 5: Post-Grant Challenges (PTAB)** - IPR, PGR, CBM proceedings - Challenge outcomes and validity assessment **Value:** Complete prosecution and challenge intelligence across patent lifecycle """ def _get_workflows_assistant_section() -> str: """Pinecone Assistant + FPD research workflows""" return """## Pinecone Assistant Integration (Optional) ### Overview Optional integration with Pinecone Assistant MCP for petition legal research. **Availability:** If Pinecone Assistant MCP is available in current session **Graceful Degradation:** All FPD workflows function fully without Assistant **Value:** Research CFR interpretations and Director policy (Free) before extracting documents **Knowledge Base:** MPEP sections, 37 CFR interpretations, Director policy memoranda ### High-Value Assistant Scenarios #### Scenario 1: Revival Petition Strategy **Problem:** Understanding 37 CFR 1.137 revival requirements **Workflow:** 1. Find similar revival petitions ```python fpd_search_petitions_balanced(petition_type_code='551', limit=20) ``` 2. Assistant research ```python assistant_context(query='37 CFR 1.137 revival requirements MPEP 711 unintentional delay', top_k=3, snippet_size=1024, temperature=0.3) ``` 3. Understand Director standards and common denial reasons 4. Select 2-3 successful petitions based on guidance 5. Extract petition text ```python fpd_get_document_content(petition_id, doc_id) ``` **Value:** Draft petition aligned with current Director standards #### Scenario 2: Examiner Dispute Analysis **Problem:** Supervisory review petition (37 CFR 1.181) - understanding standards **Workflow:** 1. Find similar disputes ```python fpd_search_petitions_balanced(petition_type_code='181', art_unit='2128', limit=20) ``` 2. Assistant research ```python assistant_context(query='37 CFR 1.181 supervisory review examiner error standards precedent', top_k=3, snippet_size=1024, temperature=0.3) ``` 3. Understand what constitutes reversible examiner error 4. Review granted petitions 5. Extract most relevant petition/decision pairs **Value:** Understand Director's standards for reversible error ### Petition Type Assistant Mapping **37 CFR 1.137 (Revival):** - Query: "37 CFR 1.137 revival unintentional delay requirements precedent" - Key Research: What constitutes 'unintentional' delay, timing, evidence needed **37 CFR 1.181 (Supervisory Review):** - Query: "37 CFR 1.181 supervisory review examiner error reversible standards" - Key Research: Reversible examiner error standards, success patterns **37 CFR 1.182 (Withdrawal/Abandonment):** - Query: "37 CFR 1.182 petition withdraw holding abandonment requirements" - Key Research: Grounds for withdrawal, procedural requirements **37 CFR 1.183 (Suspension of Rules):** - Query: "37 CFR 1.183 petition suspend rules requirements extraordinary circumstances" - Key Research: Extraordinary circumstances, evidence required ### When NOT to Use Assistant - Searching for specific petitions - use FPD search tools - Getting petition documents - use fpd_get_petition_details - Data aggregation - use FPD search tools - User already knows CFR requirements - proceed directly to analysis """ def _get_tools_section() -> str: """Tool catalog, progressive disclosure, parameters""" return """## Available Tools ### Search Tools #### 1. fpd_search_petitions_minimal **Purpose:** Ultra-fast discovery (50-100 petitions) **Context Reduction:** 95-99% **Fields:** 8 essential fields **Use When:** Initial exploration, finding petitions by company/art unit/decision type **Convenience Parameters (9 total):** - **Core Identity:** applicant_name, application_number, patent_number - **Decision Filters:** decision_type, deciding_office - **Date Ranges:** petition_date_start, petition_date_end, decision_date_start, decision_date_end **Example:** ```python fpd_search_petitions_minimal( applicant_name='TechCorp', decision_type='DENIED', limit=50 ) ``` #### 2. fpd_search_petitions_balanced **Purpose:** Detailed analysis (10-20 petitions) **Context Reduction:** 80-88% **Fields:** 18 key fields including legal context **Use When:** After minimal search, for cross-MCP analysis **Additional Parameters (5 more):** - **Petition Classification:** petition_type_code, art_unit, technology_center - **Status Filters:** prosecution_status, entity_status **Example:** ```python fpd_search_petitions_balanced( art_unit='2128', petition_type_code='551', decision_type='DENIED', limit=20 ) ``` #### 3. fpd_search_petitions_by_art_unit **Purpose:** Art unit quality assessment **Returns:** Balanced field set **Example:** ```python fpd_search_petitions_by_art_unit( art_unit='2128', date_range='2020-01-01:2024-12-31' ) ``` #### 4. fpd_search_petitions_by_application **Purpose:** Complete petition history for application **Returns:** Balanced field set **Example:** ```python fpd_search_petitions_by_application(application_number='17896175') ``` ### Detail Tools #### 5. fpd_get_petition_details **Purpose:** Full petition details by UUID **Returns:** All fields, optional documentBag with proxy URLs **Example:** ```python fpd_get_petition_details( petition_id='uuid-here', include_documents=True ) ``` #### 6. fpd_get_document_download **Purpose:** Browser-accessible PDF download URLs via secure proxy **Returns:** Proxy download URL (port 8081) **Example:** ```python fpd_get_document_download( petition_id='uuid', document_identifier='ABC123' ) ``` #### 7. fpd_get_document_content **Purpose:** Extract text from petition PDFs for LLM analysis **Extraction:** Hybrid PyPDF2 (free) + Mistral OCR (~$0.001/page) **Example:** ```python fpd_get_document_content( petition_id='uuid', document_identifier='DSEN5APWPHOENIX' ) ``` ### Progressive Disclosure Workflow **Stage 1: Discovery** - Tool: fpd_search_petitions_minimal - Volume: 50-100 results - Action: Broad search, present top results **Stage 2: Selection** - Action: User selects petitions of interest - Tip: Present 3-5 most relevant results **Stage 3: Analysis** - Tool: fpd_search_petitions_balanced - Volume: 10-20 results - Action: Detailed analysis with legal context **Stage 4: Deep Dive** - Tools: fpd_get_petition_details, fpd_get_document_download - Action: Full details and document access **Stage 5: Cross-MCP** - Tools: pfw_search_applications, ptab_search_proceedings - Action: Cross-reference with prosecution and PTAB """ def _get_red_flags_section() -> str: """Petition red flag indicators and CFR rules""" return """## Red Flag Indicators ### Revival Petitions **CFR Rule:** 37 CFR 1.137 **Meaning:** Application was abandoned, petition filed to revive **Indicator:** Missed deadlines, IP management issues, docketing problems **Workflow:** Check PFW for abandonment reason and post-revival quality **Context:** Procedural compliance issues during prosecution **Petition Type Code:** 551 ### Examiner Disputes **CFR Rule:** 37 CFR 1.181 **Meaning:** Petition for supervisory review (challenging examiner action) **Indicator:** Contentious prosecution or examiner disagreements **Workflow:** Get examiner from PFW, check if pattern with this examiner **Context:** Prosecution challenges or examination quality concerns **Petition Type Code:** 181 ### Restriction Petitions **CFR Rule:** 37 CFR 1.182 **Meaning:** Petition challenging restriction requirement **Indicator:** Claim scope issues, multiple invention issues **Workflow:** Review PFW for divisional applications and claim strategy **Context:** Claim complexity or scope issues **Petition Type Code:** 182 ### Suspension of Rules **CFR Rule:** 37 CFR 1.183 **Meaning:** Petition to suspend rules for extraordinary circumstances **Indicator:** Unusual procedural situations **Workflow:** Review petition text to understand circumstances **Context:** Special circumstances requiring Director discretion **Petition Type Code:** 183 ### Denied Petitions **Indicator:** decisionTypeCodeDescriptionText: DENIED **Meaning:** Director denied the petition **Context:** Unsuccessful arguments or procedural errors **Workflow:** Review petition details to understand why denied **Red Flag Severity:** High - indicates documented procedural problems ### Multiple Petitions **Indicator:** Same application has 2+ petitions **Meaning:** Multiple procedural issues during prosecution **Context:** Complex prosecution or persistent problems **Workflow:** Use PFW to correlate petition dates with prosecution events **Red Flag Severity:** Very High - pattern of prosecution issues ### Art Unit Red Flags **Indicator:** Art unit petition rate >15% **Calculation:** (petitions / applications) * 100 **Meaning:** Systematic art unit quality issues **Workflow:** Break down by examiner, check citation quality **Context:** May indicate training gaps or examination problems """ def _get_documents_section() -> str: """Document extraction, downloads, proxy configuration""" return """## Document Downloads and Extraction ### Always-On Proxy Configuration **Default Behavior:** Proxy server starts immediately when MCP launches - **Port:** 8081 (avoids conflict with PFW proxy on 8080) - **Startup:** Automatic - no delay or initialization needed - **Availability:** Ready instantly for all download requests - **Session:** Remains running for entire MCP session **Download Experience:** - All calls: Instant - proxy is always ready - No waiting: Download links work immediately - Reliability: High - proxy handles all authentication **Simple Workflow:** 1. Call `fpd_get_petition_details(petition_id=X, include_documents=True)` 2. Provide download links to user immediately - proxy is ready 3. Users can click links instantly ### Persistent Links (Progressive Enhancement) **Default:** generate_persistent_link=True (tries persistent links first) **Progressive Enhancement Pattern:** 1. First attempt: Try persistent link (requires USPTO PFW MCP) 2. If successful: Return 7-day encrypted persistent URL 3. If fails (no PFW): Error message suggests retry with generate_persistent_link=false 4. LLM automatically retries: Call with generate_persistent_link=false 5. Second attempt: Return immediate session-based link **Benefits:** - Users with PFW: Get persistent links automatically - Users without PFW: Graceful degradation to immediate links - No manual configuration needed ### Document Extraction **Tool:** fpd_get_document_content **Hybrid Extraction Method:** 1. Try PyPDF2 extraction (free) 2. Check extraction quality 3. If poor quality, fallback to Mistral OCR (~$0.001/page) **Cost Savings:** 70% vs OCR-only (PyPDF2 works for 60-70% of documents) **Use Cases:** - Analyze petition legal arguments and Director's reasoning - Extract CFR rules cited and statutory references - Detect patterns across multiple petitions - Correlate petition text with PTAB challenge strategies - Profile examiner behavior from supervisory review petitions **Example:** ```python fpd_get_document_content( petition_id='uuid', document_identifier='DSEN5APWPHOENIX', auto_optimize=True # Default: try PyPDF2 first ) ``` ### USPTO MCP Ecosystem Integration **FPD Standalone:** - HTTP proxy works standalone for downloads - Always-on mode provides instant links - Session-based URLs work while FPD MCP is running **FPD + PFW Together:** - PFW centralized database provides 7-day encrypted links - Links work across MCP restarts and can be shared - Unified rate limiting across all USPTO MCPs - Cross-MCP document sharing and caching **Installation Recommendation:** Install both FPD + PFW for complete patent lifecycle analysis """ def _get_ultra_context_section() -> str: """PFW fields parameter + ultra-minimal workflows""" return """## Ultra Context Reduction with PFW ### Overview PFW MCP supports ultra-context reduction (99%) via fields parameter - enables 5x broader discovery. **Benefit:** Search 100-200 results in same token budget as 20-50 results **Use Case:** Need application numbers for FPD lookup without full prosecution context ### Fields Parameter Usage **Parameter Name:** fields **Parameter Type:** List[str] **Availability:** pfw_search_applications_minimal, pfw_search_patents_minimal **Context Reduction:** - Standard minimal: 95% reduction (15 fields) - Ultra minimal: 99% reduction (2-5 fields) - Token savings: 80% additional savings vs standard minimal **Example:** ```python pfw_search_applications_minimal( examiner_name='Smith, John', fields=['applicationNumberText', 'examinerNameText'], limit=100 ) ``` **Scaling Benefit:** 100 ultra-minimal results = 20 standard minimal results (same tokens) ### Common Field Combinations **Art Unit Mapping:** ```python fields=['applicationNumberText', 'groupArtUnitNumber', 'examinerNameText'] ``` Use Case: Map art unit examiner assignments for FPD analysis **Company Identification:** ```python fields=['applicationNumberText', 'firstApplicantName', 'inventionTitle'] ``` Use Case: Identify company applications for petition lookup **Patent Identification:** ```python fields=['applicationNumberText', 'patentNumber', 'inventionTitle'] ``` Use Case: Map applications to patents for PTAB cross-reference ### FPD-Specific Workflows **Examiner Petition Analysis:** 1. Ultra-minimal discovery: ```python pfw_search_applications_minimal(examiner_name='Smith, John', fields=['applicationNumberText', 'filingDate'], limit=100) ``` 2. Extract application numbers 3. For each application: `fpd_search_petitions_by_application(application_number=app_num)` 4. Aggregate petition statistics: revival rate, dispute rate, denial rate 5. Pattern: High petition rate = examiner quality issues **Benefit:** Analyze 100 examiner applications vs 20 without fields parameter **Art Unit Petition Correlation:** 1. Get art unit applications: ```python pfw_search_applications_minimal(art_unit='2128', filing_date_start='2020-01-01', fields=['applicationNumberText', 'examinerNameText'], limit=150) ``` 2. Get art unit petitions: `fpd_search_petitions_by_art_unit(art_unit='2128')` 3. Calculate petition rate: (petitions / applications) * 100 4. Break down by examiner 5. Identify high-petition examiners **Benefit:** 150 applications analyzed vs 30 without fields parameter ### When to Use Ultra-Minimal **Use fields parameter when:** - Need application numbers for FPD petition lookup - Building examiner/art unit application lists - Company portfolio mapping for petition analysis - Large-scale discovery (100-200 results needed) - Token budget constrained but need broad coverage **Use standard minimal when:** - Need prosecution context beyond identifiers - Presenting results directly to user - Not doing cross-MCP lookup - Results will be final output """ def _get_cost_section() -> str: """Cost optimization for document extraction""" return """## Cost Optimization ### Document Extraction Costs **Overview:** Document extraction costs ~$0.001/page when Mistral OCR needed **Warning Threshold:** Alert user if extraction will exceed $0.50 **Hybrid Extraction (Default):** 1. Try PyPDF2 extraction (free) 2. Check extraction quality 3. If poor quality, fallback to Mistral OCR (~$0.001/page) **Cost Savings:** 70% vs OCR-only approach **Typical Costs:** - Text-based PDFs: $0.00 (PyPDF2 success rate: 60-70%) - Scanned PDFs: ~$0.001/page (Mistral OCR) - Most petition analyses: Well under $0.50 total ### Cost Optimization Strategy 1. **Use Research First (Free)** - Use Pinecone Assistant for CFR research (free) - Review OCR snippets in search results - Identify most relevant documents before extraction 2. **Extract Selectively** - Only extract documents needed for analysis - Start with 1-2 documents to verify relevance - Scale up only if needed 3. **Leverage PyPDF2** - Keep auto_optimize=True (default) - Let system try free extraction first - Only pay for OCR when necessary 4. **Batch Analysis** - Group similar petitions - Extract representative samples - Extrapolate patterns without extracting all ### Mistral API Key (Optional) **Required:** Only if document is scanned and PyPDF2 fails **Without Key:** Works for text-based PDFs via PyPDF2 **With Key:** Full hybrid extraction for all document types **Configuration:** Set MISTRAL_API_KEY environment variable ### Common Use Cases - Cost Examples **Due Diligence (10-20 petitions):** - Review petition metadata: $0.00 - Extract 3-5 denied petitions: $0.01-0.05 - Total: <$0.10 **Art Unit Analysis (50 petitions):** - Review petition patterns: $0.00 - Extract 5 representative samples: $0.02-0.10 - Total: <$0.15 **Examiner Profiling (100+ petitions):** - Aggregate petition statistics: $0.00 - Extract 10 dispute petitions: $0.05-0.20 - Total: <$0.25 """ def get_tool_reflections() -> str: """ DEPRECATED: Use get_guidance_section() instead. Get comprehensive guidance on FPD MCP tools and cross-MCP integration workflows. This function returns all sections concatenated for backward compatibility. Returns: Markdown-formatted string containing complete tool catalog, workflows, and integration patterns """ sections = [ _get_overview_section(), _get_workflows_pfw_section(), _get_workflows_ptab_section(), _get_workflows_citations_section(), _get_workflows_complete_section(), _get_workflows_assistant_section(), _get_tools_section(), _get_red_flags_section(), _get_documents_section(), _get_ultra_context_section(), _get_cost_section() ] return "\n\n---\n\n".join(sections) + "\n\n**End of Tool Guidance**\n"