@djm204/agent-skills 1.0.0

package/templates/professional/knowledge-synthesis/CLAUDE.md

@@ -0,0 +1,593 @@
+# Knowledge Synthesis Development Guide
+
+Principal-level guidelines for building a personal knowledge management system that ingests documents, constructs knowledge graphs, manages citations, and provides intelligent search and retrieval across your entire research corpus.
+
+---
+
+## Overview
+
+This guide applies to:
+
+- Document ingestion, parsing, and classification
+- Knowledge graph construction and maintenance
+- Citation management and bibliographic tracking
+- Note-taking systems and inter-note linking (Zettelkasten, evergreen notes)
+- Search and retrieval optimization across heterogeneous content
+- Summarization hierarchies from atomic notes to executive overviews
+- Concept mapping and ontology evolution
+- Research workflow integration
+- Version tracking of ideas over time
+
+### Key Principles
+
+1. **Atomic Knowledge Units** - Break every insight into its smallest meaningful piece
+2. **Connection Over Collection** - A note without links is a dead note; value lives in the graph
+3. **Progressive Summarization** - Layer meaning on top of raw material in successive passes
+4. **Source Fidelity** - Always preserve the link back to the original source, page, and context
+5. **Organic Growth** - Let structure emerge from content; do not force a taxonomy before you have data
+6. **Retrieval First** - Design every structure around the question "How will I find this again?"
+
+### Core Frameworks
+
+| Framework | Purpose |
+|-----------|---------|
+| Zettelkasten Method | Atomic, linked notes that form emergent structure |
+| PARA Method | Organize by actionability: Projects, Areas, Resources, Archive |
+| Progressive Summarization | Layered highlighting to surface key ideas over time |
+| Concept Mapping | Visual representation of relationships between ideas |
+| Citation Graph Analysis | Trace influence and lineage of ideas across sources |
+| Bloom's Taxonomy | Classify notes by cognitive depth: remember, understand, apply, analyze, evaluate, create |
+| Feynman Technique | Test understanding by explaining concepts in plain language |
+| Knowledge Graph Triples | Subject-Predicate-Object structure for machine-readable relationships |
+
+---
+
+## Document Ingestion and Classification
+
+### Ingestion Pipeline
+
+```text
+Document Ingestion Pipeline
+├── Intake
+│   ├── PDF extraction (academic papers, books, reports)
+│   ├── Web clipping (articles, blog posts, documentation)
+│   ├── Note capture (handwritten OCR, voice transcription)
+│   ├── Email and message extraction
+│   └── Annotation import (Hypothes.is, Kindle highlights)
+├── Parsing
+│   ├── Text extraction and cleaning
+│   ├── Metadata extraction (title, author, date, DOI)
+│   ├── Section segmentation (abstract, methods, results)
+│   ├── Figure and table extraction
+│   └── Reference list parsing
+├── Classification
+│   ├── Topic modeling (LDA, BERTopic)
+│   ├── Domain tagging (auto + manual)
+│   ├── Document type identification (paper, book, note, report)
+│   ├── Quality and relevance scoring
+│   └── Duplicate and near-duplicate detection
+├── Enrichment
+│   ├── Citation resolution and linking
+│   ├── Author disambiguation
+│   ├── Keyword extraction (TF-IDF, RAKE, KeyBERT)
+│   ├── Named entity recognition
+│   └── Cross-reference to existing knowledge base
+└── Storage
+    ├── Full-text index
+    ├── Metadata store
+    ├── Vector embeddings for semantic search
+    └── File archive with versioning
+```
+
+### Document Type Classification
+
+| Type | Characteristics | Processing Priority | Retention |
+|------|----------------|-------------------|-----------|
+| Academic Paper | DOI, abstract, citations, peer-reviewed | High | Permanent |
+| Book Chapter | ISBN, structured sections, index | High | Permanent |
+| Technical Report | Internal/external, dated, versioned | Medium | Review annually |
+| Blog Post / Article | URL, author, date, variable quality | Medium | Review bi-annually |
+| Personal Note | Self-authored, timestamped, contextual | High | Permanent |
+| Lecture / Talk | Transcript, slides, speaker | Medium | Permanent |
+| Email / Thread | Conversational, contextual, ephemeral | Low | Archive after extraction |
+| Social Media | Short-form, links, reactions | Low | Extract insights only |
+
+### Metadata Schema
+
+```text
+Document Metadata
+├── Identity
+│   ├── Unique ID (UUID)
+│   ├── Title
+│   ├── Authors []
+│   ├── Date (published / accessed)
+│   ├── DOI / ISBN / URL
+│   └── Document type
+├── Classification
+│   ├── Topics [] (hierarchical)
+│   ├── Tags [] (flat)
+│   ├── Domain
+│   ├── Relevance score (1-5)
+│   └── Quality score (1-5)
+├── Provenance
+│   ├── Source (where obtained)
+│   ├── Date ingested
+│   ├── Ingestion method (manual / automatic)
+│   └── Original file hash
+└── Processing State
+    ├── Parsed (yes/no)
+    ├── Summarized (level 0-3)
+    ├── Linked (yes/no)
+    └── Reviewed (date / never)
+```
+
+### Duplicate Detection Strategy
+
+| Method | Use Case | Precision |
+|--------|----------|-----------|
+| Exact file hash (SHA-256) | Identical files | Perfect |
+| DOI / ISBN match | Published works | Very High |
+| Title + Author fuzzy match | Near-duplicates | High |
+| MinHash / SimHash | Content similarity | Moderate |
+| Embedding cosine similarity > 0.95 | Semantic duplicates | Moderate |
+
+---
+
+## Knowledge Graph Construction
+
+### Triple Structure
+
+Every relationship follows Subject-Predicate-Object: `[Entity A] --relationship--> [Entity B]`
+
+### Relationship Taxonomy
+
+| Relationship Type | Notation | Example |
+|-------------------|----------|---------|
+| Supports / Confirms | `--supports-->` | Evidence reinforcing a claim |
+| Contradicts / Challenges | `--contradicts-->` | Opposing evidence or argument |
+| Extends / Builds Upon | `--extends-->` | Incremental development of an idea |
+| Is Instance Of | `--instance_of-->` | Specific example of a general concept |
+| Is Part Of | `--part_of-->` | Component within a larger structure |
+| Causes / Leads To | `--causes-->` | Causal relationship |
+| Is Analogous To | `--analogous_to-->` | Structural similarity across domains |
+| Supersedes / Replaces | `--supersedes-->` | Updated understanding |
+| Depends On / Requires | `--depends_on-->` | Prerequisite relationship |
+| Derived From | `--derived_from-->` | Intellectual lineage |
+
+### Graph Construction Process
+
+```text
+Graph Building Workflow
+├── 1. Extract Entities
+│   ├── Concepts and terms
+│   ├── People and organizations
+│   ├── Methods and techniques
+│   ├── Dates and events
+│   └── Claims and hypotheses
+├── 2. Identify Relationships
+│   ├── Explicit (stated in text)
+│   ├── Implicit (inferred from context)
+│   └── Cross-document (linking separate sources)
+├── 3. Validate
+│   ├── Check for contradictions
+│   ├── Verify relationship directionality
+│   ├── Confirm entity disambiguation
+│   └── Flag uncertain connections with confidence scores
+├── 4. Merge
+│   ├── Deduplicate entities
+│   ├── Resolve synonyms and aliases
+│   ├── Reconcile conflicting relationships
+│   └── Update confidence scores
+└── 5. Maintain
+    ├── Prune orphan nodes quarterly
+    ├── Review low-confidence edges
+    ├── Recompute centrality metrics
+    └── Archive superseded relationships
+```
+
+### Graph Health Metrics
+
+| Metric | Target | Warning Threshold |
+|--------|--------|-------------------|
+| Orphan nodes (no edges) | < 5% | > 15% |
+| Average node degree | 3-8 connections | < 2 or > 20 |
+| Largest connected component | > 80% of nodes | < 60% |
+| Average path length | < 5 hops | > 8 hops |
+| Clustering coefficient | 0.3 - 0.7 | < 0.1 |
+
+---
+
+## Citation Management
+
+### Citation Lifecycle
+
+```text
+Citation Workflow
+├── Capture
+│   ├── Auto-extract from PDF metadata
+│   ├── DOI lookup and enrichment (CrossRef, OpenAlex)
+│   ├── Manual entry with validation
+│   └── Import from reference managers (Zotero, Mendeley, BibTeX)
+├── Store
+│   ├── Canonical bibliographic record
+│   ├── Multiple format export (BibTeX, RIS, CSL-JSON)
+│   ├── Link to full-text file
+│   └── Link to all notes referencing this source
+├── Use
+│   ├── Inline citation insertion
+│   ├── Bibliography generation
+│   ├── Citation context extraction (surrounding sentences)
+│   └── Citation intent classification (background, method, result, comparison)
+└── Analyze
+    ├── Most-cited sources in personal corpus
+    ├── Citation clusters and communities
+    ├── Temporal citation patterns
+    └── Missing citations (referenced but not captured)
+```
+
+### Citation Intent Classification
+
+| Intent | Description | Example Signal |
+|--------|-------------|---------------|
+| Background | General context or motivation | "It is well known that..." |
+| Uses / Method | Adopting a technique or tool | "Following the approach of..." |
+| Comparison | Contrasting results or methods | "Unlike [Author], we find..." |
+| Extension | Building on prior work | "Extending the framework of..." |
+| Critique | Disagreeing or identifying flaws | "However, [Author] overlooks..." |
+| Support | Providing evidence for a claim | "Consistent with [Author]..." |
+
+### Bibliographic Record Fields
+
+Every record must include: unique key (e.g., Smith2024), title, authors, year, venue, DOI/ISBN/URL, document type, tags, date added, date read, relevance score (1-5), 1-2 sentence summary, key quotes with page numbers, and links to derived Zettelkasten notes.
+
+---
+
+## Note-Taking Systems and Linking
+
+### Zettelkasten Implementation
+
+```text
+Zettelkasten Structure
+├── Fleeting Notes
+│   ├── Quick captures, unprocessed
+│   ├── Inbox for raw thoughts
+│   └── Must be processed within 48 hours or archived
+├── Literature Notes
+│   ├── Paraphrased ideas from a single source
+│   ├── Always linked to bibliographic record
+│   ├── Written in your own words (never copy-paste)
+│   └── One note per key idea (not per source)
+├── Permanent Notes
+│   ├── Atomic: one idea per note
+│   ├── Self-contained: understandable without context
+│   ├── Connected: linked to related permanent notes
+│   ├── Written for your future self
+│   └── Includes source reference
+├── Structure Notes (Maps of Content)
+│   ├── Index notes that organize clusters of permanent notes
+│   ├── Provide narrative through a topic
+│   ├── Updated as new notes are added
+│   └── Multiple structure notes can reference the same permanent note
+└── Project Notes
+    ├── Tied to a specific output (paper, presentation, project)
+    ├── Temporary: archived when project completes
+    └── Reference permanent notes, do not duplicate them
+```
+
+### Note Quality Checklist
+
+Before a note earns "permanent" status:
+
+- [ ] Contains exactly one idea
+- [ ] Written in complete sentences (not bullet fragments)
+- [ ] Understandable without reading the source
+- [ ] Has at least one link to another permanent note
+- [ ] Includes source reference with page/section
+- [ ] Tagged with relevant topics
+- [ ] Unique ID assigned (timestamp-based recommended: YYYYMMDDHHMMSS)
+
+### Link Types Within Notes
+
+| Link Type | Purpose | Notation Example |
+|-----------|---------|-----------------|
+| Direct Reference | "This note elaborates on..." | `[[202601011200]]` |
+| See Also | Related but not dependent | `See also: [[202512150900]]` |
+| Contradicts | Opposing viewpoint | `Contrast with: [[202511201400]]` |
+| Sequence | Part of a chain of reasoning | `Next: [[202601021000]]` |
+| Structure | Belongs to a map of content | `Part of: [[MOC-Machine-Learning]]` |
+| Source | Derived from this document | `Source: [[Smith2024]]` |
+
+### Progressive Summarization Layers
+
+| Layer | Action | Result |
+|-------|--------|--------|
+| 0 - Capture | Save the full source text | Raw material in archive |
+| 1 - Bold | Bold the most important passages | ~10-20% of original highlighted |
+| 2 - Highlight | Highlight within the bolded passages | ~2-5% of original stands out |
+| 3 - Executive Summary | Write a 2-3 sentence summary in your words | Distilled insight |
+| 4 - Remix | Combine with other notes into new output | Original contribution |
+
+---
+
+## Search and Retrieval Optimization
+
+### Search Architecture
+
+```text
+Search Stack
+├── Lexical Search (BM25 / Full-Text)
+│   ├── Exact keyword matching
+│   ├── Boolean operators (AND, OR, NOT)
+│   ├── Field-specific search (title, author, tags)
+│   └── Best for: known-item retrieval, exact phrases
+├── Semantic Search (Vector / Embedding)
+│   ├── Meaning-based similarity
+│   ├── Handles synonyms and paraphrases
+│   ├── Cross-language retrieval
+│   └── Best for: exploratory search, "notes about X concept"
+├── Graph Traversal
+│   ├── Follow relationship edges
+│   ├── Find shortest paths between concepts
+│   ├── Cluster discovery
+│   └── Best for: "How does A relate to B?", serendipitous discovery
+└── Hybrid (Reciprocal Rank Fusion)
+    ├── Combine lexical + semantic scores
+    ├── Re-rank with cross-encoder
+    ├── Filter by metadata facets
+    └── Best for: general-purpose queries
+```
+
+### Query Patterns
+
+| Query Type | Example | Best Search Method |
+|------------|---------|-------------------|
+| Known item | "Find the Smith 2024 paper on transformers" | Lexical (title + author) |
+| Topical | "Everything I have on spaced repetition" | Hybrid (semantic + tag filter) |
+| Relational | "What contradicts my note on X?" | Graph traversal |
+| Temporal | "What did I read about AI safety last quarter?" | Lexical + date filter |
+| Exploratory | "Surprise me with something related to this idea" | Semantic + random walk |
+| Synthesis | "What are the main arguments for and against Y?" | Graph + semantic retrieval |
+
+### Embedding Strategy
+
+| Content Type | Embedding Granularity | Model Recommendation |
+|-------------|----------------------|---------------------|
+| Academic papers | Per-paragraph | Domain-specific (SciBERT, SPECTER) |
+| Book chapters | Per-section | General-purpose (text-embedding-3-large) |
+| Personal notes | Per-note (atomic) | General-purpose |
+| Highlights | Per-highlight | General-purpose |
+| Metadata | Per-record | Lightweight (text-embedding-3-small) |
+
+### Retrieval Quality Metrics
+
+| Metric | Definition | Target |
+|--------|-----------|--------|
+| Precision@10 | Relevant results in top 10 | > 0.7 |
+| Recall@50 | Fraction of all relevant docs found in top 50 | > 0.8 |
+| Mean Reciprocal Rank | Average rank of first relevant result | > 0.6 |
+| User satisfaction (self-rated) | Did you find what you needed? | > 4/5 |
+| Time to answer | Seconds from query to useful result | < 30s |
+
+---
+
+## Summarization Hierarchies
+
+### Multi-Level Summarization
+
+```text
+Summarization Pyramid
+├── Level 0: Raw Source
+│   └── Full original text, unmodified
+├── Level 1: Extracted Notes
+│   └── Key passages, quotes, data points (per source)
+├── Level 2: Atomic Summaries
+│   └── One-paragraph summary per concept/claim
+├── Level 3: Topic Summaries
+│   └── Multi-source synthesis per topic (1-2 pages)
+├── Level 4: Domain Overviews
+│   └── Comprehensive survey of a research area (5-10 pages)
+└── Level 5: Executive Brief
+    └── Cross-domain strategic summary (1 page)
+```
+
+### Summarization Quality Criteria
+
+| Criterion | Description | Verification Method |
+|-----------|-------------|-------------------|
+| Accuracy | No factual errors or misrepresentations | Compare against source |
+| Completeness | All key claims preserved | Check coverage of main points |
+| Conciseness | No redundancy or filler | Word count within target range |
+| Attribution | Every claim traceable to a source | Verify citation links |
+| Neutrality | Does not inject opinion unless labeled | Review for editorializing |
+| Coherence | Logical flow, no non sequiturs | Read aloud test |
+
+### Topic Summary Template
+
+```text
+## Topic Summary: [Topic Name]
+
+**Last Updated**: [YYYY-MM-DD]
+**Sources Synthesized**: [Count]
+**Confidence Level**: [High / Medium / Low]
+
+### Definition
+[What is this topic? 2-3 sentences.]
+
+### Key Claims
+1. [Claim] — supported by [Source1], [Source2]
+2. [Claim] — supported by [Source3]; challenged by [Source4]
+3. [Claim] — emerging evidence from [Source5]
+
+### Open Questions
+- [Question that remains unresolved]
+- [Contradiction between sources]
+
+### Related Topics
+- [[Topic A]] — [nature of relationship]
+- [[Topic B]] — [nature of relationship]
+
+### Timeline of Understanding
+| Date | Development | Source |
+|------|------------|--------|
+| [YYYY-MM] | [How understanding evolved] | [Source] |
+```
+
+---
+
+## Research Workflow Integration
+
+### Research Phases and Knowledge Actions
+
+| Phase | Knowledge Action | System Support |
+|-------|-----------------|---------------|
+| Question Formation | Search existing notes for prior thinking | Semantic search, graph browse |
+| Literature Review | Ingest and classify new sources | Ingestion pipeline, dedup |
+| Reading and Annotation | Create literature notes, highlight | Progressive summarization |
+| Analysis | Create permanent notes, link to graph | Zettelkasten workflow |
+| Synthesis | Build topic summaries, concept maps | Summarization hierarchy |
+| Writing | Retrieve and cite from knowledge base | Citation manager, search |
+| Revision | Update notes based on new understanding | Version tracking |
+| Publication | Export bibliography, archive project notes | Export, archival |
+
+### Integration Points
+
+| Tool / System | Integration Method | Data Flow |
+|--------------|-------------------|-----------|
+| Reference Manager (Zotero) | BibTeX / CSL-JSON sync | Bidirectional |
+| PDF Reader (Zotero, PDF Expert) | Annotation export | One-way into system |
+| Note App (Obsidian, Logseq) | Markdown files, backlinks | Bidirectional |
+| Search Engine (Elasticsearch, Meilisearch) | Index on change | One-way from system |
+| Vector DB (Qdrant, Weaviate, Chroma) | Embed on note create/update | One-way from system |
+| Writing Tool (LaTeX, Google Docs) | Citation insertion plugin | One-way from system |
+| Calendar / Task Manager | Research task scheduling | Metadata sync |
+| Web Browser | Clip extension | One-way into system |
+
+---
+
+## Version Tracking of Ideas
+
+### Idea Evolution Model
+
+```text
+Idea Lifecycle
+├── Seed
+│   ├── First encounter with a concept
+│   ├── Minimal understanding, high uncertainty
+│   └── Captured as fleeting or literature note
+├── Sprout
+│   ├── Connected to 2-3 other ideas
+│   ├── Paraphrased in own words
+│   └── Promoted to permanent note
+├── Growth
+│   ├── Refined through multiple encounters
+│   ├── Evidence accumulates (supporting and opposing)
+│   └── Featured in a structure note or concept map
+├── Maturity
+│   ├── Well-connected, well-evidenced
+│   ├── Summarized at multiple levels
+│   └── Used in original output (writing, presentations)
+├── Revision
+│   ├── Challenged by new evidence
+│   ├── Updated with version annotation
+│   └── Old version preserved, not deleted
+└── Archival / Supersession
+    ├── Replaced by a more complete understanding
+    ├── Linked to successor note
+    └── Marked as superseded with date and reason
+```
+
+### Version Annotation Format
+
+```text
+## Note: [Title] (v3)
+
+**Created**: 2025-06-15
+**Last Updated**: 2026-01-28
+**Version**: 3
+**Status**: Active / Revised / Superseded
+
+### Current Understanding
+[The current state of this idea]
+
+### Change Log
+| Version | Date | Change | Trigger |
+|---------|------|--------|---------|
+| v1 | 2025-06-15 | Initial capture from [Source] | Reading |
+| v2 | 2025-09-03 | Refined after encountering [Source2] | Contradiction |
+| v3 | 2026-01-28 | Expanded with evidence from [Source3] | New evidence |
+
+### Previous Versions
+- v2: [[202509031400-note-title-v2]]
+- v1: [[202506151200-note-title-v1]]
+```
+
+### Tracking Intellectual Lineage
+
+| Question | Method |
+|----------|--------|
+| Where did this idea originate? | Follow `derived_from` edges backward |
+| How has my thinking changed? | Compare version change logs |
+| What evidence changed my mind? | Review "Trigger" column in change log |
+| Which ideas are most stable? | Sort by version count ascending (fewer revisions = more stable) |
+| Which ideas are most contested? | Count `contradicts` edges |
+| What are my intellectual blind spots? | Find domains with few notes or low connectivity |
+
+---
+
+## Common Pitfalls
+
+### 1. Collector's Fallacy
+
+Wrong: Save hundreds of PDFs, highlight obsessively, never synthesize. Hoarding feels like learning but produces no understanding.
+
+Right: For every hour of reading, spend 30 minutes writing notes in your own words. If you cannot explain it, you have not learned it.
+
+### 2. Over-Engineering Taxonomy
+
+Wrong: Spend weeks designing the perfect folder hierarchy and tagging system before adding a single note.
+
+Right: Start with minimal structure. Let tags and links emerge organically. Reorganize only when retrieval fails.
+
+### 3. Orphan Notes
+
+Wrong: Create notes but never link them. Each note is an island.
+
+Right: Every new note must connect to at least one existing note. If you cannot find a connection, question whether the note belongs in your system.
+
+### 4. Copy-Paste Without Processing
+
+Wrong: Highlight a passage and paste it directly into your notes.
+
+Right: Paraphrase in your own words. The act of translation is the act of understanding. Preserve the original only as a reference quote.
+
+### 5. Ignoring Contradictions
+
+Wrong: When two sources disagree, keep both without comment.
+
+Right: Create explicit contradiction links. Write a note exploring the tension. Contradictions are where the most valuable thinking happens.
+
+### 6. Neglecting Maintenance
+
+Wrong: Build the system for six months, then never prune, review, or update.
+
+Right: Schedule weekly reviews. Prune dead notes quarterly. Update summaries when new evidence arrives. A knowledge system is a garden, not a warehouse.
+
+### 7. Tool Obsession
+
+Wrong: Spend more time configuring plugins, themes, and workflows than actually reading and thinking.
+
+Right: Pick a tool that supports backlinks and full-text search. Start writing. Optimize only when you hit a real bottleneck.
+
+---
+
+## Resources
+
+- [How to Take Smart Notes - Sonke Ahrens](https://www.goodreads.com/book/show/34507927-how-to-take-smart-notes)
+- [Building a Second Brain - Tiago Forte](https://www.goodreads.com/book/show/59616977-building-a-second-brain)
+- [Zettelkasten.de](https://zettelkasten.de/)
+- [Andy Matuschak's Evergreen Notes](https://notes.andymatuschak.org/)
+- [Obsidian](https://obsidian.md/) / [Logseq](https://logseq.com/)
+- [Zotero Reference Manager](https://www.zotero.org/)
+- [Semantic Scholar API](https://api.semanticscholar.org/)
+- [OpenAlex API](https://docs.openalex.org/)
+- [Connected Papers](https://www.connectedpapers.com/)

package/templates/professional/life-logistics/.cursor/rules/financial-optimization.mdc

@@ -0,0 +1,78 @@
+---
+description: Personal Finance Optimization
+alwaysApply: false
+---
+
+# Personal Finance Optimization
+
+Guidelines for reducing recurring costs, optimizing insurance plans, maximizing rewards and benefits, and maintaining a systematic approach to household financial management.
+
+## Optimization Principles
+
+1. **Measure before optimizing** - Collect at least 6-12 months of spending data before recommending changes
+2. **Focus on the big categories first** - Housing, insurance, transportation, and utilities yield the largest savings
+3. **Optimization is recurring** - A one-time review is not enough. Financial optimization requires quarterly and annual review cadences
+4. **Tax-advantaged accounts first** - Maximize HSA, FSA, and retirement account contributions before optimizing discretionary spending
+5. **Rewards stacking** - Layer credit card rewards, loyalty programs, employer benefits, and promotional offers for maximum value
+
+## Insurance Optimization Methodology
+
+### Usage Pattern Analysis
+
+Before comparing any insurance plan:
+
+1. Pull 24 months of claims history and out-of-pocket spending
+2. List every provider, medication, and service used with frequency
+3. Identify any planned procedures or changes in the next 12 months
+4. Verify which providers are in-network for each plan under consideration
+5. Calculate total annual cost under best-case, expected-case, and worst-case usage scenarios
+
+### Plan Comparison
+
+For every insurance comparison, calculate total annual cost as:
+
+- (Monthly Premium x 12) + Expected Out-of-Pocket Based on Usage - Tax Savings from HSA/FSA - Employer Contributions
+
+Weight the scenarios: 50% best case, 35% expected case, 15% worst case. The plan with the lowest weighted total annual cost is typically the best choice, subject to network adequacy.
+
+### Annual Insurance Shopping Schedule
+
+- **45-60 days before renewal**: Request current policy declarations page and begin competitor quoting
+- **30-45 days before renewal**: Complete comparisons, negotiate with current carrier using competitor quotes
+- **14-30 days before renewal**: Make final decision and bind new policy if switching
+- **Day of renewal**: Verify new policy is active and old policy terminates without gap
+
+## Expense Category Strategies
+
+### High-Impact Categories
+
+- **Housing**: Review refinance opportunities when rates drop 0.5%+ below current rate. Appeal property tax assessments when comparable sales support a lower valuation. Request PMI removal at 80% LTV.
+- **Insurance**: Shop all policies annually. Bundle home and auto for 10-25% discount. Optimize deductibles based on emergency fund size.
+- **Telecom**: Negotiate annually with retention department. Right-size data and channel plans to actual usage.
+- **Subscriptions**: Audit quarterly. Cancel anything unused in the past 30 days. Negotiate annual rates for keepers.
+
+### Credit Card Rewards
+
+- Assign spending categories to the card with the highest reward rate for that category
+- Set recurring bills on the card that offers the best rate for subscription or utility payments
+- Review reward program changes quarterly as issuers frequently modify category bonuses
+- Redeem rewards at their highest-value option (often travel or statement credit, not merchandise)
+
+## Financial Review Cadence
+
+Follow the review schedule in CLAUDE.md:
+
+- **Weekly**: Transaction review for unauthorized charges, verify autopay processing
+- **Monthly**: Budget reconciliation, upcoming renewal check for the next 60 days
+- **Quarterly**: Subscription audit, insurance adequacy review, rewards optimization check
+- **Annually**: Full insurance shopping, tax withholding review, provider rate comparison, credit report check
+- **Event-triggered**: Any rate increase, life change, claim, or contract expiration
+
+## Verification Protocol
+
+After any financial change (new rate, plan switch, account modification):
+
+1. Document the change with confirmation number, effective date, and expected impact
+2. Set a 30-day reminder to verify the change on the next statement
+3. If the change is not reflected, escalate immediately with documentation
+4. Update the renewal tracker with the new terms and next review date