softaworks/agent-toolkit▌

Create professional Marp presentation slides with 7 pre-designed themes and automatic quality enhancements. \n \n Includes 7 built-in themes (default, minimal, colorful, dark, gradient, tech, business) with embedded CSS, each optimized for different content types and audiences \n Provides structured templates and best-practice guidelines for slide layout, typography, and visual hierarchy \n Supports image integration using Marp syntax for side layouts, full backgrounds, and multi-image compositi

Detect and remove AI writing patterns to make text sound naturally human. \n \n Identifies 24 distinct AI-writing patterns including inflated symbolism, promotional language, vague attributions, em dash overuse, rule of three, AI vocabulary words, and superficial -ing analyses \n Rewrites problematic sections while preserving meaning, tone, and voice, with guidance on adding personality and specificity to avoid soulless writing \n Based on Wikipedia's comprehensive \"Signs of AI writing\" guide,

Generate software architecture documentation using C4 model Mermaid diagrams. \n \n Supports all four C4 levels: System Context, Container, Component, and Deployment diagrams, plus Dynamic diagrams for request flows \n Includes syntax and examples for people, systems, containers, components, boundaries, and relationships with directional and styling options \n Provides microservices modeling guidance for single-team and multi-team ownership, plus event-driven architecture patterns \n Recommends

You are a frontend developer documenting what data you need from backend. You describe the what, not the how. Backend owns implementation details.

React useEffect best practices guide covering when to use Effects and superior alternatives. \n \n Teaches the escape-hatch nature of Effects: use only for synchronizing with external systems, not for derived state, expensive calculations, or user event responses \n Provides a decision tree and quick-reference table mapping common situations (data fetching, state derivation, prop changes) to the correct React pattern \n Covers when NOT to use Effects: transforming data for render, handling user

Generate memes using 100+ templates from the memegen.link API with custom text and styling. \n \n Supports popular templates like Drake, Buzz Lightyear, Success, and Fine, each suited to specific messaging contexts (comparisons, celebrations, irony, uncertainty) \n Text encoding handles spaces, newlines, special characters, and quotes through simple character substitution ( _ for spaces, ~q for question marks, etc.) \n Includes dimension and layout options for platform-specific sizing (1200x630

Seamless AI agent session transfers with comprehensive context preservation and validation. \n \n Automatically triggers on user requests, context window capacity warnings, task milestones, or work session endings; proactively suggests handoffs after substantial work \n CREATE workflow generates timestamped handoff documents with pre-filled project metadata, git history, and modified files, then validates for completeness and security \n RESUME workflow loads handoffs with staleness checking, co

Strunk's timeless rules for clearer, stronger prose in documentation, commit messages, and UI text. \n \n Covers 18 core principles from The Elements of Style including active voice, positive form, concrete language, and ruthless concision \n Includes reference files on grammar, composition, formatting, and common word misuse; most tasks need only the composition section (~4,500 tokens) \n Identifies and helps avoid AI writing patterns: puffery, empty \"-ing\" phrases, promotional adjectives, an

Use ONLY when user says "search", "find", "look up", "ask", "research", or "what's the latest" for generic queries. NOT for library/framework docs (use Context7), gt CLI (use Graphite MCP), or workspace questions (use Nx MCP).

Templates and guidance for writing READMEs matched to your project type and audience. \n \n Identifies four project types (Open Source, Personal, Internal, Config) with distinct audience needs and required sections \n Provides a structured three-step process: identify the task (creating, adding, updating, or reviewing), ask task-specific questions, then solicit final feedback \n Includes essential baseline sections for all READMEs: name, description, and usage with examples \n References support

Core principle: Main agents NEVER read Excalidraw files directly. Always delegate to subagents to isolate context consumption.

For diagrams used in Quarto slides, specify defaultFontFamily in mxGraphModel tag:

$22

Create logical, well-described git commits with staged review and Conventional Commits formatting. \n \n Guides you through inspecting changes, deciding commit boundaries, and staging only intended modifications using patch mode when needed \n Enforces Conventional Commits format (type, scope, subject, body, footer) with clear separation of what changed and why \n Includes a pre-commit checklist covering secrets detection, accidental debug code, and unrelated formatting to catch issues before sh

Type-safe React 18–19 patterns with compile-time guarantees for components, hooks, events, and routing. \n \n Covers React 19 breaking changes: ref as prop (no forwardRef), useActionState for forms, use() for promise unwrapping, and Server Components with async data fetching \n Specific event typing (MouseEvent, ChangeEvent, KeyboardEvent, etc.) and discriminated unions for variant props eliminate runtime surprises \n Generic component patterns with constrained types for reusable Tables, Lists,

Multi-step planning orchestration: research, stakeholder interviews, specification synthesis, implementation planning, and external LLM review. \n \n Guides users through structured discovery: research topics, detailed interviews, and spec synthesis before writing implementation plans \n Generates self-contained section files with dependencies, requirements, and acceptance criteria for autonomous or manual implementation \n Integrates feedback from external LLM reviewers (Gemini, Codex) and supp

AI-powered code analysis, refactoring, and automated editing via Codex CLI with GPT-5.2. \n \n Runs codex exec and codex resume commands with configurable reasoning effort ( xhigh , high , medium , low ) and sandbox modes (read-only, workspace-write, danger-full-access) \n Defaults to GPT-5.2 model (76.3% SWE-bench performance); supports gpt-5.2-max, gpt-5.2-mini, and gpt-5.1-thinking for different complexity and cost trade-offs \n Session continuity: resume prior Codex work at any time using co

Guides creation of reusable Claude Code slash commands for automating repetitive workflows. \n \n Helps determine command location (project-level .claude/commands/ or global ~/.claude/commands/ ) and walks through gathering command name, purpose, arguments, and workflow steps \n Includes four command patterns: workflow automation, iterative fixing, agent delegation, and simple execution \n Provides bundled references covering patterns, real examples, best practices, and quality checklists to ens

Analyze code naming conventions and suggest clearer, more consistent variable, function, and class names. \n \n Identifies issues across variables, functions, classes, files, database objects, and API endpoints including vague names, misleading behavior, abbreviations, and convention violations \n Checks language-specific conventions (camelCase, snake_case, PascalCase) and framework patterns with special attention to boolean naming prefixes \n Provides prioritized suggestions with reasoning, fro

A CLI tool for AI agents to debug and triage using Datadog logs and metrics.

Material-UI v7 (released March 2025) patterns for component usage, styling with sx prop, theme integration, and responsive design.

Convert web pages to clean Markdown by driving a locally installed browser (via web2md).

No Chat Output: Produce the handoff document only. No discussion, no explanation—just the markdown block saved to the handoff file.

Automatically transforms vague requirements into actionable PRDs through systematic clarification with a 100-point scoring system.

Natural language Jira interaction with automatic backend detection and safety guardrails. \n \n Supports two backends: Jira CLI (if installed locally) and Atlassian MCP tools, with automatic detection and fallback guidance \n Covers core workflows: viewing, creating, updating, and transitioning issues; listing sprints and personal tickets; assigning and commenting \n Includes built-in safety checks: always fetches current state before modifying, shows changes for approval, verifies transitions a

Structured framework for preparing, delivering, and following up on challenging workplace conversations. \n \n Three-phase approach covering preparation (clarify issue, check emotions, consider perspective), delivery (open neutrally, share perspective, listen actively, seek resolution), and followup (document agreements, check progress, maintain relationship) \n Includes the SBI model (Situation-Behavior-Impact) for framing feedback and guidance on managing your own emotions before difficult con

Structured frameworks for navigating difficult conversations and delivering constructive feedback effectively. \n \n Two core models: Preparation-Delivery-Follow-up (three-phase conversation structure) and Situation-Behavior-Impact (SBI) feedback technique that focuses on observable facts rather than assumptions \n Covers preparation strategies including issue analysis, goal definition, and emotional regulation to reduce defensiveness and improve outcomes \n Delivery phase includes neutral openi

Generate comprehensive test plans, test cases, regression suites, and bug reports with Figma design validation. \n \n Create structured test plans with scope, strategy, entry/exit criteria, risk assessment, and timelines \n Generate manual test cases with step-by-step instructions, expected results, preconditions, and priority levels \n Build regression test suites (smoke, targeted, full) with execution order and pass/fail criteria \n Validate UI implementations against Figma designs with compon

Generate creative domain names and check real-time availability across multiple TLDs. \n \n Analyzes your project description to create relevant, memorable domain suggestions across .com, .io, .dev, .ai, .app, and other extensions \n Verifies actual availability for each suggestion and flags taken domains with estimated pricing \n Supports advanced workflows including variation checking, industry-specific naming, and competitor pattern analysis \n Provides branding rationale for each suggestion

Production-ready SQL and NoSQL schemas with normalization, indexing, and migration strategies built in. \n \n Generates complete schemas from natural language descriptions, covering entities, relationships, constraints, and indexes tailored to SQL or NoSQL databases \n Provides normalization guidance (1NF through 3NF), foreign key strategies with ON DELETE rules, and composite index planning based on access patterns \n Includes anti-patterns reference (FLOAT for money, missing FK indexes, non-re

Minimize total codebase size by systematically identifying and removing unnecessary code. \n \n Focuses on final code amount, not effort or churn; a 50-line addition that deletes 200 lines is a win \n Requires loading a reference mindset from the skill's philosophy directory before proceeding \n Applies three core questions: what's the smallest codebase that solves this, does the change reduce total code, and what can be deleted as a result \n Flags common traps like status quo bias, premature f

Text-based software diagrams for domain modeling, system architecture, data flows, and process visualization. \n \n Supports 9+ diagram types including class diagrams, sequence diagrams, flowcharts, entity relationship diagrams, C4 architecture diagrams, state machines, git graphs, and charts \n All diagrams defined as plain text with version-controllable .mmd files; native rendering in GitHub, GitLab, VS Code, Notion, and Obsidian \n Includes detailed references for each diagram type covering s

Frameworks and best practices for clear, professional communication across emails, team chat, meetings, and technical audiences. \n \n Covers four core areas: the What-Why-How structure for organizing messages, email templates with subject line formulas, team messaging etiquette including the \"no hello\" principle, and strategies for translating technical concepts to non-technical audiences \n Includes audience calibration guidance, jargon-to-plain-language translation examples, and clarity pri

Transform learning content into concrete, shippable implementation cycles using the Ship-Learn-Next framework. \n \n Converts passive content (transcripts, articles, tutorials) into actionable rep-based plans with specific weekly goals and success criteria \n Structures learning as repeating cycles: Ship (create something real), Learn (reflect honestly), Next (iterate based on insights) \n Emphasizes doing over studying—each rep produces a tangible artifact and builds one new skill, designed to

Convert OpenAPI 3.0 specifications to TypeScript interfaces and type guards. \n \n Accepts OpenAPI files in JSON or YAML format and validates against OpenAPI 3.0.x specification \n Generates TypeScript interfaces from components/schemas , request/response types from paths , and runtime type guards for validation \n Handles complex types including objects, arrays, enums, unions (oneOf), and inheritance (allOf) with proper required/optional field mapping \n Includes automatic JSDoc comments from O

Strategic product thinking to identify game-changing features and 10x opportunities. \n \n Guides structured analysis across three scales: massive (transformative), medium (high-leverage), and small (disproportionate value) opportunities \n Provides a ruthless evaluation framework assessing impact, reach, frequency, differentiation, defensibility, and feasibility for each idea \n Includes nine idea categories (speed, automation, intelligence, integration, collaboration, personalization, visibili

Interactive daily standup generator that pulls activity from GitHub, Jira, and Claude Code history. \n \n Detects available integrations (GitHub CLI, Jira, Claude Code session history) and asks permission before pulling any data \n Conducts a structured 4-question interview (yesterday's work, today's plan, blockers, discussion topics) with pulled data as context to trigger memory \n Generates clean Markdown updates combining interview responses with tool data, including PR links and ticket summa

Scaffold, structure, and manage Claude Code plugins with automated manifest generation and version control. \n \n Includes Python scripts for creating plugins and bumping semantic versions across both plugin.json and marketplace.json manifests \n Supports four component types: commands (with namespace-based naming), skills, agents, and hooks, each with defined directory structures \n Provides three plugin patterns (framework, utility, domain) with reference documentation covering structure, mark

Build robust, scalable design systems that ensure visual consistency and exceptional user experiences.

Extract software engineering lessons from recent git changes by analyzing actual code patterns. \n \n Analyzes code diffs against a curated principles reference to identify dominant patterns, trade-offs, and structural decisions \n Supports multiple scopes: feature branches vs. main, specific commits, commit ranges, or uncommitted working changes \n Maps findings to concrete SE principles with specific file and line references, avoiding generic advice \n Presents lessons in a reflective format t

Multi-language dependency management with auto-detection, safe PATCH/MINOR updates, and prompted MAJOR version changes. \n \n Supports 7 languages (Node.js, Python, Go, Rust, Ruby, Java, .NET) with language-specific package managers and audit tools \n Automatically categorizes updates by semver type: auto-applies PATCH and MINOR, prompts individually for MAJOR versions, skips intentionally pinned dependencies \n Includes diagnosis mode for resolving version conflicts, peer dependency issues, and

Refactor bloated agent instruction files into organized, linked documentation using progressive disclosure. \n \n Analyzes existing AGENTS.md, CLAUDE.md, or similar files to identify contradictions, extract essentials, and categorize remaining instructions across 3–8 focused topic files \n Keeps root file minimal (under 50 lines) with only universal project info and links to detailed guidelines organized by topic (TypeScript, testing, code style, git workflow, architecture) \n Flags redundant, v

Comprehensive code review and large-context analysis using Gemini 3 Pro with flexible approval modes. \n \n Supports five Gemini models ranging from flagship reasoning (Gemini 3 Pro) to cost-optimized alternatives, with 1M token input windows for analyzing entire codebases and documentation sets \n Three approval modes: default for interactive terminal prompts, auto_edit for auto-approving code suggestions, and yolo for fully automated background execution \n Critical safety requirement: use --a