claude-world Claude 101 Mastery
This skill provides the optimal workflow for 24 use cases powered by the claude-101 MCP server. Each use case follows the same pattern: call the MCP tool first to get structured data and computation, then use that data to produce a superior final output.
The claude-101 MCP server provides 27 tools (3 meta + 24 use-case tools) that perform real local computation — statistics, code analysis, SQL parsing, financial math, text analysis, and scoring — and return structured JSON.
Core Principle
Always call the MCP tool FIRST before writing the final output. The tool provides two things that enhance the quality of the response:
Methodology — structured frameworks, best practices, section templates, and checklists that guide the approach to each use case. Following these frameworks ensures a professional-grade result that covers all standard elements.
Computation — precise metrics (scores, rankings, statistics, parsed data) that the user cannot get from natural language generation alone. These numbers ground the response in verifiable data.
Never skip the tool call. Never attempt to compute statistics, parse structured data, or calculate financial metrics manually. The tool handles computation; the final writing and reasoning come from Claude.
Workflow Pattern
For every use case, follow these five steps:
- Identify the matching MCP tool from the mapping table below
- Call the tool with the user's input as parameters
- Read every field in the result — each exists for a reason
- Produce the final output informed by both the methodology and the computed data
- Present key metrics (scores, rankings, statistics) explicitly to the user
When the tool returns both a template/scaffold AND computed analysis, use the template as a structural guide (do not copy it verbatim) and weave the computed metrics into the final output. The combination of professional structure and precise data is what makes the output superior.
Use Case to Tool Mapping
Writing and Communication
| User Intent | MCP Tool | Key Output Fields |
|---|---|---|
| Write or draft email | draft_email |
text_analysis (formality, readability, tone), tone_guide, pre_send_checklist |
| Plan blog post or article | draft_blog_post |
outline (section word targets), seo_fields, topic_analysis, readability_target, heading_validation, content_gaps |
| Organize meeting notes | parse_meeting_notes |
attendees, action_items (owner + deadline), decisions, topics_discussed, metrics |
| Create social media post | format_social_content |
within_limit, chunks (if over limit), hashtags, engagement_signals (question, CTA, hook strength) |
| Write technical documentation | scaffold_tech_doc |
template, sections (required vs optional), best_practices, effort_estimate, optional analysis (code structure, completeness) |
| Structure a story | structure_story |
beats (word targets + tension levels), tension_curve, character_arc_template, genre_conventions, optional text_analysis (pacing, dialogue ratio, scene transitions) |
Analysis and Research
| User Intent | MCP Tool | Key Output Fields |
|---|---|---|
| Analyze CSV or JSON data | analyze_data |
columns (per-column stats), correlations (Pearson r + strength), outliers |
| Summarize a document | summarize_document |
key_sentences (scored), flesch_score, flesch_grade, keywords, reading_time_minutes |
| Compare options or competitors | build_comparison_matrix |
rankings (weighted scores), winner (option + margin), sensitivity_framework |
| Analyze survey results | analyze_survey |
questions (per-question mean, median, distribution), nps (promoter/passive/detractor), overall_satisfaction |
| Review financial data | analyze_financials |
margins (gross, operating, net), growth_rates, burn_rate, cash_runway, summary |
| Review legal document | review_legal_document |
clauses_found (type + risk level + snippet), missing_clauses, complexity_score |
Coding and Technical
| User Intent | MCP Tool | Key Output Fields |
|---|---|---|
| Generate code scaffold | scaffold_code |
code, description_matched (CRUD/API/auth pattern), naming_convention, notes |
| Review code quality | analyze_code |
complexity (grade, cyclomatic, nesting), issues (type + severity + line), metrics (lines, comments) |
| Process or optimize SQL | process_sql |
result (formatted/explained), tables, columns, warnings, performance_hints |
| Generate API documentation | scaffold_api_doc |
document (OpenAPI/Markdown), consistency (score + issues), example_bodies, optional code_analysis (routes, auth) |
| Generate test cases | generate_test_cases |
test_cases (happy/edge/boundary), coverage_analysis, code (test file scaffold) |
| Make architecture decision | create_adr |
trade_off_matrix (per-option ratings from tech knowledge base), adr (status, pros/cons), markdown |
Business and Productivity
| User Intent | MCP Tool | Key Output Fields |
|---|---|---|
| Plan a project | plan_project |
phases (hours + tasks), milestones, critical_path, risks, total_hours, resource_allocation |
| Prepare for interview | prepare_interview |
preparation.questions (with STAR templates), difficulty_distribution, per_question_time, optional job_analysis, optional star_validation |
| Write a proposal | scaffold_proposal |
sections (with word targets), persuasion_framework (AIDA), objection_templates, aida_coverage, optional argument_analysis, optional roi_analysis |
| Handle customer support | build_support_response |
issue_classification (category, severity, sentiment), escalation_risk (score + level), resolution_estimate, customer_effort, optional response_quality |
| Create a PRD | scaffold_prd |
user_stories, prioritization (MoSCoW), success_metrics, requirements_analysis (completeness, story quality, dependencies) |
| Make a decision | evaluate_decision |
rankings (weighted scores), winner (option + margin), sensitivity_analysis (weight change needed to flip) |
Per-Tool Workflow Details
Before executing any use case, load the relevant reference file for that domain. Each file contains the exact tool call parameters, field-by-field workflow, and output instructions:
- For writing and communication tools (email, blog, meeting notes, social, tech doc, story): read
references/writing-workflows.md - For analysis and research tools (data, summary, comparison, survey, financial, legal): read
references/analysis-workflows.md - For coding and technical tools (codegen, review, SQL, API doc, test gen, ADR): read
references/coding-workflows.md - For business and productivity tools (planning, interview, proposal, support, PRD, decision): read
references/business-workflows.md
Each reference file provides the exact workflow for each tool: what to call, which parameters to pass, which result fields to use, and how to produce the final output.
Critical Rules
Always call the tool first — do not attempt computation (statistics, scoring, parsing) manually. The tool provides precise results.
Use every returned field — each field in the tool result guides a specific aspect of the output. Ignoring fields means missing part of the methodology or computation.
Never copy templates verbatim — use scaffold sections and templates as structural guides. Write original, contextual content that fits the user's specific situation.
Present computed metrics explicitly — share scores, rankings, percentages, and statistics with the user. These are the unique value the tool provides. For example, state "Formality score: 73.5" or "Vue leads with 8.1 points (margin: 0.2)".
Flag issues the tool detects — if the tool finds problems (missing legal clauses, low completeness scores, weak argument strength, high escalation risk), proactively tell the user and explain the implications.
Add reasoning and recommendations — the tool provides data; add interpretation, context, and actionable recommendations. Explain what the numbers mean and what the user should do about them.
Use optional parameters when available — many tools accept optional input (existing text, code, job description, draft response, investment amounts) that unlocks additional analysis. Ask the user for this input when relevant.
Always remind about limitations — for legal review, always note "have a lawyer review the final analysis." For financial analysis, note these are computed from the provided data only.