62 agent skills for reasoning, context engineering, and professional work. Works with Claude Code, Cursor, Codex, and 40+ agents.
# All skills (any agent)
npx skills add AndurilCode/craftwork
# List available skills first
npx skills add AndurilCode/craftwork --list
# Specific skills only
npx skills add AndurilCode/craftwork --skill reasoning-orchestrator --skill agent-instruction-forge
# Specific agent only
npx skills add AndurilCode/craftwork -a claude-code# Add the marketplace
/plugin marketplace add https://github.com/AndurilCode/craftwork
# Install by category
/plugin install craftwork-reasoning@craftwork # 23 thinking frameworks (includes orchestrator)
/plugin install craftwork-context-engineering@craftwork # 13 context/agent tools (includes orchestrator)
/plugin install craftwork-professional@craftwork # 26 architecture/communication skills (includes orchestrator)
/plugin install craftwork-all@craftwork # everything (62 skills)Thinking frameworks for analysis, decisions, and problem-solving. Includes its own orchestrator for guided routing.
| Category | Skills |
|---|---|
| Understand the System | systems-thinking, first-principles-thinking, cynefin-framework, epistemic-mapping |
| Find What's Broken | theory-of-constraints, five-whys-root-cause, causal-inference, cognitive-bias-detection |
| Stress-Test | inversion-premortem, red-teaming, second-order-thinking, limit-thinking |
| Navigate Uncertainty | probabilistic-thinking, fermi-estimation, scenario-planning, game-theoretic-analysis |
| Generate Options | lateral-thinking, analogical-thinking, bisociative-creativity |
| Decide | decision-synthesis, decision-intelligence, retrospective-counterfactual, evidence-synthesis |
| Orchestrate | reasoning-orchestrator — entry point, triages and routes to the right framework |
Build, evaluate, and debug agent context — instructions, harnesses, evals, and documentation. Includes its own orchestrator for guided routing.
| Skill | Purpose |
|---|---|
| agent-instruction-forge | Create instruction rules for coding agents (CLAUDE.md, AGENTS.md, .cursorrules) |
| rule-quality-evaluator | Score instruction rules on Seven Properties, detect redundancies, test behaviorally |
| context-cartography | Design what goes into an agent's context window |
| context-gap-analyzer | Find implicit context missing from a codebase |
| context-debugging | Diagnose agent failures that originate in the context layer |
| context-eval | Measure whether a context harness actually improves outcomes |
| edd | Eval-Driven Development — TDD for context, not code |
| llms-txt-generator | Generate token-efficient context documents for LLM consumption |
| deep-document-processor | Multi-pass reading to extract decision-relevant context from large documents |
| business-logic-extractor | Extract domain rules and business logic from code into structured references |
| context-compressor | Maximize signal-per-token under a finite context budget |
| test-challenger | Find false positives in AI-generated tests |
| Orchestrate | context-engineering-orchestrator — entry point, routes to the right context skill |
Architecture, code quality, process design, communication, and leadership. Includes its own orchestrator for guided routing.
| Skill | Purpose |
|---|---|
| architecture-evaluation | Evaluate system design decisions and produce ADRs |
| code-review-amplifier | Amplify human code reviewers with structured pre-scanning |
| debugging-methodology | Systematic debugging: reproduce, observe, hypothesize, isolate |
| execution-planning | Decompose decisions into executable plans with dependencies |
| experimental-design | Design rigorous experiments to validate assumptions |
| process-design | Design or redesign workflows using Lean and value stream analysis |
| financial-modeling | Unit economics, cost-benefit analysis, NPV/IRR, scenario modeling |
| argument-craft | Structure recommendations into persuasive arguments |
| narrative-construction | Turn analysis into compelling stories |
| negotiation-strategy | Prepare for negotiations with BATNA, ZOPA, and concession planning |
| difficult-conversations | Navigate conflict, feedback, and emotionally charged discussions |
| facilitation-design | Design meetings and workshops that produce decisions |
| stakeholder-power-mapping | Map influence networks and design engagement strategies |
| ethical-reasoning | Surface moral implications using multiple ethical frameworks |
| fairness-auditing | Audit systems for equitable outcomes across groups |
| learning-strategy | Build structured plans for closing knowledge gaps |
| casual-inference | Distinguish causation from correlation in metrics and experiments |
| technical-writing | Write RFCs, design docs, ADRs, runbooks, postmortems, one-pagers, announcements |
| topic-explainer | Explain concepts, technologies, or ideas using the best style for the topic |
| summarizer | Summarize documents, articles, transcripts, or multi-source content |
| presentation-craft | Create presentation scripts with narrative arc, slide visuals, and speaker notes |
| kaizen | Continuous improvement audit: find waste, unevenness, and overburden in code |
| kintsugi | Repair visibility audit: find undocumented fixes and add context gold |
| knowledge-architect | Capture decisions, context, and learnings; design team knowledge systems |
| skill-router | Exhaustive skill scan and composition planning for any request |
| Orchestrate | professional-orchestrator — entry point, routes to the right professional skill |
Skills compose. Common sequences:
Evaluating something — Cynefin → Systems Thinking → Theory of Constraints → 5 Whys → Causal Inference
Validating a plan — Epistemic Mapping → First Principles → Inversion/Pre-mortem → Second-Order → Decision Synthesis
High-stakes decision — Scenario Planning → Probabilistic Thinking → Fermi Estimation → Red Teaming → Decision Synthesis
Stuck with no options — Epistemic Mapping → Lateral Thinking → Analogical Thinking → First Principles
Scaling decision — Limit Thinking → Second-Order Thinking → Scenario Planning → Decision Synthesis
After something goes wrong — Retrospective/Counterfactual → 5 Whys → Epistemic Mapping → Decision Synthesis
Full context engineering lifecycle — Context Gap Analyzer → Agent Instruction Forge → Rule Quality Evaluator → Context Eval → EDD
Creating agent instructions — Context Gap Analyzer (audit) → Agent Instruction Forge (create) → Rule Quality Evaluator (score) → EDD (validate)
Architecture decision — Architecture Evaluation → Argument Craft → Execution Planning
Navigating organizational resistance — Stakeholder Power Mapping → Negotiation Strategy → Difficult Conversations
Building a business case — Financial Modeling → Argument Craft → Execution Planning
Each plugin group has its own orchestrator that routes within its skills. When multiple groups are installed, cross-group chains are documented in routing.yaml.
skills/ # flat — npx skills discovers these
├── reasoning-orchestrator/
├── context-engineering-orchestrator/
├── professional-orchestrator/
├── first-principles-thinking/
├── agent-instruction-forge/
├── context-eval/
├── ...62 skills total
│ └── SKILL.md
plugins/ # Claude Code marketplace plugins
├── craftwork-reasoning/ → 23 skills (includes orchestrator)
├── craftwork-context-engineering/ → 13 skills (includes orchestrator)
├── craftwork-professional/ → 26 skills (includes orchestrator)
└── craftwork-all/ → 62 skills (all orchestrators)
routing.yaml # single source of truth for skill composition
scripts/validate-routing.sh # validates routing.yaml against actual skills
.claude-plugin/
└── marketplace.json # craftwork marketplace
Skills live flat at root for npx skills compatibility. The plugins/ directory groups skills by category for the Claude Code marketplace.
MIT