Skip to content

tachyon-beep/skillpacks

Claude Code Skill Packs Marketplace

Professional AI/ML, Python & Rust engineering, web backend, DevOps, SDLC, solution architecture, game development, security, documentation, and UX skills for Claude Code

44 complete skillpacks • 200+ skills • Install what you need

Recent additions (May 2026): 6 new packs — yzmir-morphogenetic-rl, axiom-determinism-and-replay, axiom-audit-pipelines, axiom-static-analysis-engineering, axiom-rust-workspaces, axiom-pyo3-interop. Plus a 2026-era refresh sweep across the AI/ML cluster (PyTorch 2.9+, FSDP2, vLLM/SGLang, GRPO/DreamerV3, Mamba/SSM, modern PEFT) and the design/security/quality packs (WCAG 2.2, Material 3, Pydantic v2, Typst 0.14, SLSA/SBOM/Sigstore).

In-flight expansions: lyra-creative-writing v0.2 — genre annex with 9 genre sheets (mystery, thriller, sf, fantasy, horror, romance, literary-fiction, memoir-and-personal-essay, literary-journalism) using a reader-contract three-move discipline, plus 3 new agents (worldbuilding-consultant, opening-and-ending-doctor, premise-stress-tester). axiom-system-archaeologist v1.6 — ultralarge-tier per-module track for repos >100K LOC or >12 subsystems, with calibration-driven YAML-parse self-validation.


Installation

Via Marketplace (Recommended)

# Add the skillpacks marketplace
/plugin marketplace add tachyon-beep/skillpacks

# Browse available plugins
/plugin

# Install specific packs
/plugin install yzmir-deep-rl
/plugin install lyra-ux-designer
/plugin install ordis-security-architect

Via Git Clone (Development)

git clone https://github.com/tachyon-beep/skillpacks
cd skillpacks
/plugin marketplace add .

Available Skillpacks

Skillpack groupings are inspired by the factions from Altered TCG. See FACTIONS.md for the philosophical connections between factions and skillpack domains.

🔒 Security (Ordis)

ordis-security-architect - 11 skills, 3 commands, 2 agents (refreshed for 2026)

  • Threat modeling (STRIDE / CWE / CVSS / ATT&CK), security controls
  • Compliance: NIST CSF 2.0, ISO 27001:2022, PCI-DSS v4.0.1, GDPR, NIS2, EU AI Act
  • LLM/AI security (OWASP LLM Top 10:2025, MITRE ATLAS), supply-chain (SLSA, SBOM, Sigstore), ATO/RMF
  • /plugin install ordis-security-architect

📝 Documentation (Muna) - 4 Packs

muna-document-designer - 1 skill, 1 agent, 1 command

  • Professional document design with Pandoc and Typst
  • Typography, layout, templates, branded document systems
  • Reports, proposals, specifications, resumes, brochures
  • /plugin install muna-document-designer

muna-technical-writer - 10 skills, 4 commands, 3 agents

  • Documentation structure, clarity & style, diagram conventions
  • Security-aware docs, incident response, ITIL/governance
  • Includes complex-writer / complex-reviewer agent pair for surgical edits in large files (≥2000 lines), cross-language
  • /plugin install muna-technical-writer

muna-panel-review - 3 agents, 1 skill, 3 commands

  • Simulated audience panel review with persona-driven editorial intelligence
  • Spawn reader personas, collect mood journals, synthesize cross-panel feedback
  • /plugin install muna-panel-review

muna-wiki-management - 7 skills, 4 commands, 2 agents

  • Document set management as wikis with architecture and governance
  • Derivation tracking, consistency auditing, evolution management
  • /plugin install muna-wiki-management

🔬 Development (Axiom) - 16 Packs

axiom-python-engineering - 10 skills

  • Modern Python 3.13: uv-first tooling, ty/pyrefly, current pre-commit revs
  • Testing, async, debugging, profiling, delinting
  • Scientific computing (NumPy/pandas), ML workflows
  • /plugin install axiom-python-engineering

axiom-rust-engineering - router + 11 reference sheets, 5 commands, 3 agents

  • Rust 2024 edition: ownership, traits, async (tokio), testing
  • Clippy/cargo tooling, performance, unsafe/FFI, AI/ML interop (PyO3, candle)
  • Single-crate-shaped — composes with axiom-rust-workspaces and axiom-pyo3-interop
  • /plugin install axiom-rust-engineering

axiom-rust-workspaces - router + 13 sheets, 3 commands, 1 agent (new — May 2026)

  • Rust at workspace scope: multi-crate composition, [workspace.dependencies], [workspace.lints], deny.toml with waiver lifecycle
  • Feature unification (resolver-2/3), Miri-on-subset, public-vs-internal crate boundaries (internal-traits-crate / sealed-trait), 10-pattern anti-pattern list
  • /plugin install axiom-rust-workspaces

axiom-pyo3-interop - router + 13 sheets, 3 commands, 1 agent (new — May 2026)

  • Python ↔ Rust FFI discipline: Bound<'py, T>, GIL release, abi3, maturin
  • Batched FFI to amortise crossing cost, zero-copy NumPy buffer protocol, Gymnasium environments backed by Rust, async across the boundary
  • Wheel matrix (cibuildwheel, manylinux), interpreter-teardown discipline
  • /plugin install axiom-pyo3-interop

axiom-system-archaeologist - 5 commands, 4 agents (v1.6 ultralarge-tier — May 2026)

  • Deep codebase architecture analysis through subagent coordination
  • C4 diagrams (Context, Container, Component levels), subsystem catalogs, security surface mapping, test infrastructure analysis, dependency visualization
  • Incremental delta analysis, cross-pack specialist integration
  • Ultralarge-tier per-module track for repos >100K LOC or >12 subsystems, with calibration-driven YAML-parse self-validation (v1.6.0 + v1.6.1)
  • Stakeholder-ready architecture documentation with output-contract validation gates
  • /plugin install axiom-system-archaeologist

axiom-system-architect - 4 skills

  • TDD-validated architectural assessment with professional discipline
  • Prevents diplomatic softening, analysis paralysis, security compromise
  • Router + 3 specialist skills for technical debt and architecture review
  • /plugin install axiom-system-architect

axiom-solution-architect - 9 skills, 2 agents, 2 commands

  • Forward solution architecture: brief/HLD/epic → traceable artifact set
  • ADRs, C4, NFRs, RTM, integration/migration, risks, TOGAF/ArchiMate
  • Consolidated Solution Architecture Document (SAD) with consistency gate
  • /plugin install axiom-solution-architect

axiom-web-backend - 12 skills

  • FastAPI, Django, Express.js development patterns
  • REST/GraphQL API design, microservices architecture
  • Authentication, database integration, message queues
  • API testing, documentation, production deployment
  • /plugin install axiom-web-backend

axiom-devops-engineering - 1 skill

  • CI/CD pipeline architecture, deployment strategies
  • Zero-downtime deployments, infrastructure reliability
  • /plugin install axiom-devops-engineering

axiom-sdlc-engineering - 8 skills + 4 agents

  • CMMI-based SDLC framework (Levels 2-4) with GitHub/Azure DevOps integration
  • Requirements lifecycle, design/build, quality assurance, governance & risk
  • Platform integration, quantitative management, lifecycle adoption
  • Specialist agents for architecture decisions, bug triage, QA, and process routing
  • /plugin install axiom-sdlc-engineering

axiom-planning - 2 skills, 5 agents, 1 command

  • TDD-validated implementation planning with plan review quality gate
  • Plan review spawns 4 parallel reviewers (reality, architecture, quality, systems) + synthesizer
  • Catches hallucinations, convention violations, and risks before code execution
  • /plugin install axiom-planning

axiom-program-management - router + 13 sheets, 3 commands, 2 agents (new — June 2026)

  • Project & program management as a delivery discipline — lean/agile-leaning, honest about where program scale needs predictive structure
  • Project tier: flow metrics & WIP, scope/backlog control, Monte-Carlo forecasting, stakeholder management, operational RAID logs, watermelon-resistant status reporting, dependency coordination
  • Program tier: governance, benefits realization & OKRs, now/next/later roadmapping with WSJF, cross-project dependencies, capacity & stable teams, scaling without framework dogma
  • Leans on axiom-planning (implementation plans) and axiom-sdlc-engineering (process maturity, traceability, formal governance)
  • /plugin install axiom-program-management

axiom-product-management - router + 8 sheets, 3 commands, 2 agents (new — June 2026)

  • Product management as standing ownership — take control of a software product and own it end-to-end across sessions
  • A git-versioned product workspace (vision, intent-roadmap, Product Decision Records, current-state, falsifiable metrics) + a resume→decide→dispatch→accept→ checkpoint loop + an authority boundary that escalates anything irreversible
  • Discovery & opportunity, vision/strategy/positioning, PRDs with falsifiable acceptance criteria, delivery orchestration & acceptance, product-value metrics
  • Orchestrates, doesn't reimplement: routes sequencing/WSJF to axiom-program-management, plans to axiom-planning, solution design to axiom-solution-architect, research method to lyra-ux-designer
  • /plugin install axiom-product-management

axiom-embedded-database - router + 13 sheets, 3 commands, 2 agents

  • SQLite & DuckDB as application-embedded databases at production scale
  • PRAGMA discipline, migrations under SQLite's ALTER constraints, WAL tuning, transactions & isolation, single-writer concurrency, optimistic locking
  • Parameterized-SQL enforcement, JSON1, FTS5, DuckDB for OLAP, SQLCipher, online backup & corruption recovery
  • /plugin install axiom-embedded-database

axiom-procedural-architecture - router + 13 sheets, 3 commands, 2 agents

  • Structural reasoning for staged procedures — design, critique, and analyze decompositions into composable stages with explicit dependencies, decision points, and flow properties
  • Pattern pack for wizards, troubleshooting trees, training curricula, configuration flows, decision pipelines, and workflows
  • /plugin install axiom-procedural-architecture

axiom-mcp-engineering - router skill (early access)

  • Engineering MCP (Model Context Protocol) servers as production-grade agent interfaces — tool API surface design, idempotency & atomic claim semantics under agent retry, structured error envelopes, schema versioning, transport reliability
  • /plugin install axiom-mcp-engineering

axiom-engineering-foundations - 6 skills

  • Universal software engineering methodology (language-agnostic)
  • Systematic debugging, safe refactoring, code review, incident response
  • Technical debt triage, codebase confidence building
  • /plugin install axiom-engineering-foundations

axiom-audit-pipelines - router + 11 sheets, 3 commands, 2 agents (new — May 2026)

  • Audit-grade decision pipelines: canonical encoding (RFC 8785 JCS), append-only decision logs, fingerprint chains (linked-hash and Merkle)
  • HMAC and Ed25519 signed exports, immutable storage, decision provenance (inputs/ruleset/code closure), threat model for the log itself
  • Retention reconciled with right-to-be-forgotten, partial replay, performance budgets
  • /plugin install axiom-audit-pipelines

axiom-determinism-and-replay - router + 13 sheets, 3 commands, 2 agents (new — May 2026)

  • Architecture-level determinism and replay for RL substrates, multi-agent systems, multiplayer lockstep engines, replay-debuggable services
  • Seed governance, RNG isolation, snapshot strategy, divergence detection, rollback, concurrency strategies, floating-point and GPU determinism
  • External-effects substitution, canonical state encoding, property tests, cost-of-determinism accounting
  • /plugin install axiom-determinism-and-replay

axiom-static-analysis-engineering - router + 13 sheets, 3 commands, 2 agents (new — May 2026, v0.2)

  • Building static analyzers as engines, not running them as users
  • AST visitation, taint-lattice (abstract-domain) design with monotonicity, three-phase inference (variable → function → callgraph) with termination proofs
  • Plugin architecture for rules, false-positive economics with auditable suppression lifecycle, the static-vs-runtime boundary
  • Callgraph construction (Rung 0–4 resolution, dynamic-feature handling), cross-module flow with stub libraries, decorator-as-assertion (runtime + static agreement contract), manifest-driven configuration with coherence validation, SARIF / CI integration, incremental analysis at scale, LLM-assisted finding explanations as a substitutable pattern
  • /plugin install axiom-static-analysis-engineering

🤖 AI/ML Engineering (Yzmir) - 11 Packs

yzmir-ai-engineering-expert - 1 router skill

  • Primary router that directs to specialized AI/ML packs
  • /plugin install yzmir-ai-engineering-expert

yzmir-pytorch-engineering - 9 skills (refreshed for PyTorch 2.9+)

  • Tensors, modules, distributed training, profiling, debugging
  • torch.compile, FSDP1/FSDP2, FlexAttention, torch.amp
  • /plugin install yzmir-pytorch-engineering

yzmir-training-optimization - 11 skills (refreshed)

  • Modern optimizers, schedules, mixed-precision, batch-size strategy
  • Learning rates, convergence, hyperparameter tuning
  • /plugin install yzmir-training-optimization

yzmir-deep-rl - 13 skills (refreshed)

  • DQN, PPO, SAC, reward shaping, exploration
  • Modern offline RL, GRPO, DreamerV3 / TD-MPC2, REDQ/DroQ/CrossQ, MAPPO/IPPO, BBF/Agent57, Go-Explore
  • /plugin install yzmir-deep-rl

yzmir-morphogenetic-rl - router + 7 sheets, 2 commands, 2 agents (new — May 2026)

  • RL controllers that decide WHEN/HOW to mutate a network's topology during training
  • Controller action/observation/reward design, governor and safety gates
  • Rollback-as-RL-signal, deterministic morphogenesis, ablation under topology change
  • Companion to yzmir-dynamic-architectures
  • /plugin install yzmir-morphogenetic-rl

yzmir-neural-architectures - 9 skills (refreshed)

  • CNNs, Transformers, RNNs, attention mechanisms, architecture selection
  • Mamba/SSM, MoE (Mixtral/DeepSeek/OLMoE), modern diffusion (SDXL/FLUX/DiT/SD3), multimodal (CLIP/SigLIP/LLaVA), SAM/SAM-2, ConvNeXt v2, equivariant GNNs
  • /plugin install yzmir-neural-architectures

yzmir-llm-specialist - 8 skills (refreshed)

  • Reasoning models, agentic + MCP, modern serving stack, DPO/GRPO
  • RAG with contextual retrieval, fine-tuning, inference optimization
  • /plugin install yzmir-llm-specialist

yzmir-ml-production - 11 skills (refreshed)

  • torch.ao.quantization, vLLM/SGLang/TensorRT-LLM serving
  • LLM observability, MLOps, monitoring, debugging
  • /plugin install yzmir-ml-production

yzmir-simulation-foundations - 9 skills

  • Differential equations, stability analysis, control theory (game math)
  • /plugin install yzmir-simulation-foundations

yzmir-systems-thinking - 6 skills

  • Systems thinking methodology, patterns, leverage points
  • Archetypes, modeling, visualization
  • /plugin install yzmir-systems-thinking

yzmir-dynamic-architectures - 6 skills, 1 agent, 2 commands (refreshed)

  • Dynamic/morphogenetic neural networks: grow, prune, adapt topology
  • Continual learning, gradient isolation, lifecycle orchestration
  • Modern PEFT (VeRA/PiSSA/LoftQ/LoRA+/rsLoRA/LongLoRA), post-Mixtral MoE, adapter merging (TIES/DARE/SLERP/MergeKit)
  • /plugin install yzmir-dynamic-architectures

🎮 Game Development (Bravos) - 2 Packs

bravos-simulation-tactics - 11 skills

  • Physics simulation, ecosystem simulation, crowd simulation
  • /plugin install bravos-simulation-tactics

bravos-systems-as-experience - 9 skills

  • Emergent gameplay, player-driven narratives, strategic depth
  • /plugin install bravos-systems-as-experience

🎨 UX, Site Design & Creative Writing (Lyra) - 3 Packs

lyra-creative-writing - 1 router + 13 craft + 9 genre sheets, 3 commands, 11 agents (v0.2 genre annex — May 2026)

  • Workshop-voiced craft pack for prose narrative — fiction and creative nonfiction
  • Three explicit modes (draft, critique, plan) with mode-switching protocol
  • Craft coverage: POV and voice, scene construction, dialogue, character interiority, showing-vs-telling, prose rhythm, pacing and tension, story structure, openings and endings, worldbuilding by implication, research and verisimilitude, revision and cutting, creative nonfiction craft
  • Genre coverage (v0.2): mystery, thriller, sf, fantasy, horror, romance, literary-fiction, memoir-and-personal-essay, literary-journalism — each presenting conventions through a uniform reader-contract three-move discipline (name the contract → name the cost of breaking → name books worth the cost). Literary fiction is one mode among nine, not the unmarked default
  • /critique-prose fans out 5 coach-mode agents in parallel and synthesises a prioritised revision-pass plan via revision-coach
  • v0.2 agents: worldbuilding-consultant (world abstraction), opening-and-ending-doctor (promise-ledger output, first-5/last-5 pages), premise-stress-tester (adversarial pre-outline interrogation)
  • Composes with muna-panel-review for beta-reader simulation
  • /plugin install lyra-creative-writing

lyra-site-designer - 1 skill, 1 agent, 1 command, 6 reference sheets

  • Static site design for developer tools and documentation sites
  • Information architecture, HTML/CSS craftsmanship, design tokens
  • Developer UX patterns: code blocks, dark mode, search, responsive layouts
  • Static site tooling guidance (Hugo, Astro, Eleventy)
  • /plugin install lyra-site-designer

lyra-ux-designer - 12 reference sheets (refreshed for 2026)

  • WCAG 2.2 accessibility, modern web platform primitives (dialog, popover, :has(), container queries)
  • iOS 17+ and Material 3 mobile patterns, AI-experience patterns
  • /plugin install lyra-ux-designer

✅ Quality Engineering (Ordis)

ordis-quality-engineering - 21 skills

  • E2E testing, API testing, integration testing, performance testing
  • Chaos engineering, flaky test diagnosis, mutation testing
  • Observability, test automation architecture, coverage gap analysis
  • /plugin install ordis-quality-engineering

🛠️ Meta Packs

meta-sme-protocol - 1 skill

  • SME (Subject Matter Expert) Agent Protocol
  • Mandatory protocol for all specialist agents: fact-finding, confidence/risk assessment
  • /plugin install meta-sme-protocol

meta-skillpack-maintenance - 3 skills

  • Systematic maintenance and enhancement of skill packs
  • Investigative domain analysis, RED-GREEN-REFACTOR testing
  • Automated quality improvements
  • /plugin install meta-skillpack-maintenance

Quick Examples

AI/ML Engineering

/plugin install yzmir-deep-rl
/deep-rl  # Use slash command to load router skill

Claude will guide you to the right specialized skill:

I'm using yzmir/deep-rl/policy-gradient-methods to implement PPO for continuous character control

Web Backend Development

/plugin install axiom-web-backend
/web-backend  # Use slash command to load router
I'm using axiom/web-backend/fastapi-development to build production-ready async APIs with dependency injection

Game Development

/plugin install bravos-systems-as-experience
/systems-as-experience  # Use slash command to load router
I'm using bravos/systems-as-experience/emergent-gameplay-design to create systemic interactions in my RPG

Security Architecture

/plugin install ordis-security-architect
/security-architect  # Use slash command to load router
I'm using ordis/security-architect/threat-modeling to analyze this authentication system with STRIDE

UX Design

/plugin install lyra-ux-designer
/ux-designer  # Use slash command to load router
I'm using lyra/ux-designer/accessibility-and-inclusive-design to ensure WCAG 2.1 AA compliance

Architecture Analysis

/plugin install axiom-system-archaeologist
/system-archaeologist  # Use slash command to load router
Then Claude routes you to specialized analysis skills for C4 diagrams, subsystem catalogs, etc.

Using Router Skills: Slash Commands vs Direct Invocation

Slash Commands (Recommended)

All router skills (using-X skills) are available as slash commands to avoid skill context limits:

/ai-engineering          # Route to AI/ML specialized packs
/deep-rl                # Route to RL algorithm skills
/system-archaeologist   # Route to architecture analysis
/python-engineering     # Route to Python skills
/web-backend           # Route to web backend skills
/ux-designer           # Route to UX design skills
# ... and 12 more

Why slash commands? Router skills are comprehensive guides that exceeded the context budget for automatic skill discovery. Slash commands provide explicit, user-controlled invocation without context limits.

See .claude/SLASH_COMMANDS.md for the complete list of all router commands.

Direct Skill Invocation

If you have only a few plugins installed, direct skill invocation still works:

I'm using yzmir/deep-rl/policy-gradient-methods to implement PPO

Caveat: With many plugins installed (5+), you may hit skill discovery context limits. In this case, use slash commands for routers and direct invocation for specialized skills.

Best Practice:

  • Use slash commands for router skills (/ai-engineering, /deep-rl, etc.)
  • Use direct invocation for specialized skills after the router guides you

What's Inside

  • 200+ Skills: Production-ready guidance across 7 domains
  • 38 Packs: Install only what you need
  • Per-Pack Plugins: Independent versioning, clean dependencies
  • Complete Coverage: Design → Foundation → Training → Production

Repository Structure

skillpacks/
├── .claude-plugin/
│   └── marketplace.json       # Marketplace catalog
├── plugins/                   # All 38 skillpacks
│   ├── axiom-*/               # Development (15 packs)
│   ├── bravos-*/              # Game development (2 packs)
│   ├── lyra-*/                # UX & site design (2 packs)
│   ├── meta-*/                # Meta utilities (2 packs)
│   ├── muna-*/                # Documentation (4 packs)
│   ├── ordis-*/               # Quality & security (2 packs)
│   └── yzmir-*/               # AI/ML (11 packs)
└── docs/                      # Planning documents

Contributing

See CONTRIBUTING.md for guidelines on adding skills or creating new packs.


License

CC BY-SA 4.0 (Creative Commons Attribution-ShareAlike 4.0 International) - See LICENSE for details.

Note: Faction names (Axiom, Bravos, Lyra, Muna, Ordis, Yzmir) from Altered TCG are NOT covered by this license - see LICENSE_ADDENDUM.md.


About

Built by @tachyon-beep

Skillpacks provides modular, production-ready expertise for Claude Code across professional domains. Each skillpack is independently versioned and installable, containing systematically validated skills that guide Claude through expert-level implementations.

About

No description, website, or topics provided.

Resources

License

Code of conduct

Stars

Watchers

Forks

Packages

 
 
 

Contributors