ITSalt

Code review for completed tasks (BE, FE, or TECH). Use when: review code, code review, check implementation, verify task, approve development, or the user says "/nacl-tl-review UC### --be" or "/nacl-tl-review UC### --fe". Flags: --be for backend review, --fe for frontend review, no flag for TECH tasks.

Code review for completed tasks (BE, FE, or TECH). Use when: review code, code review, check implementation, verify task, approve development, or the user says "/nacl:tl-review UC### --be" or "/nacl:tl-review UC### --fe". Flags: --be for backend review, --fe for frontend review, no flag for TECH tasks.

Review NaCl TL implementation against task specs, API contracts, tests, stubs, and quality checklists. Use when reviewing backend, frontend, or full UC/TECH implementation, requesting changes, approving work, or when the user says `/nacl-tl-review`.

Initialize a new project with CLAUDE.md containing development rules, skill routing, bug fix protocol, and documentation discipline. Use when: new project, initialize project, setup project, create CLAUDE.md, or the user says "/nacl-init".

Infrastructure and TECH task development using TDD workflow. Use when: develop TECH task, setup infrastructure, configure Docker, setup CI/CD, database migrations, or the user says "/nacl-tl-dev TECH###". Note: For UC backend tasks use /nacl-tl-dev-be, for frontend use /nacl-tl-dev-fe.

Spec-first bug fixing with automatic documentation sync. Auto-detects affected UC/TECH from problem description. Classifies fix level (L0/L1/L2/L3), updates docs BEFORE code. Use when: fix bug, resolve error, debug issue, something broke, or the user says "/nacl-tl-fix" followed by a problem description.

Initialize a new project with CLAUDE.md containing development rules, skill routing, bug fix protocol, and documentation discipline. Use when: new project, initialize project, setup project, create CLAUDE.md, or the user says "/nacl:init".

Infrastructure and TECH task development using TDD workflow. Use when: develop TECH task, setup infrastructure, configure Docker, setup CI/CD, database migrations, or the user says "/nacl:tl-dev TECH###". Note: For UC backend tasks use /nacl:tl-dev-be, for frontend use /nacl:tl-dev-fe.

This skill instructs Claude to perform distribution-faithful sampling and diverse generation by seeding randomness from a self-generated string rather than relying on token-completion patterns. Apply it when the task requires genuine stochasticity or meaningful variation across runs: creative writing, brainstorming, variant generation, probabilistic sampling, mixed-strategy games, and agent simulation. Trigger phrases include: "write 5 different", "brainstorm", "flip a coin", "pick randomly", "generate variants", "surprise me", "diverse", "different versions". Do not apply for math, factual lookup, code debugging, translation, classification, extraction, or any single-answer task where variability reduces correctness.

Transforms unstructured user task descriptions into production-ready prompts following the CRAFT+ methodology (classical CRAFT framework extended with 2026 best practices), with model-specific formatting for Claude, GPT, Gemini, DeepSeek, and Universal targets. Use this skill ONLY when the user explicitly asks to build, compose, write, generate, or improve a prompt for an AI model. Trigger phrases include "build a prompt", "compose a prompt", "write a prompt", "create a prompt", "improve this prompt", "fix this prompt", "make a prompt for [Claude/GPT/Gemini/DeepSeek]", "собери промпт", "составь промпт", "напиши промпт", "улучши промпт", "перепиши промпт", "сделай промпт", "промпт под [Claude/ChatGPT/Gemini/DeepSeek]". Do not activate for direct task execution — only for prompt construction.