rayk/lucid-toolkit
7 skills0 installs21 starsGitHub
Install
npx skills add https://github.com/rayk/lucid-toolkitSkills in this repo
1Analystanalyst is a journey-wide agent skill that equips solo builders with structured thinking—fourteen mental models, multi-agent problem solving, and fact-checking—whenever a decision is still fuzzy. Its canonical home in the catalog is Idea and research because that is where you first stress-test opportunities, assumptions, and competitive story before validation spend. The same toolkit remains useful when you renegotiate scope in validate, prioritize launch risks in ship, or interpret growth experiments later, because it optimizes for clarity and verified claims rather than code generation. Install it when chat brainstorming leaves you with slogans instead of traceable reasoning, or when you want multiple agent perspectives cross-checked before you invoke planning or implementation skills. It is methodology, not a single API integration.0installs2ArchitectArchitect is an agent skill plugin for Rayk’s lucid-toolkit that helps solo and indie builders design and document software using Lucid Composite Architecture (LCA). Instead of one-off diagrams, it organizes work around a structural hierarchy—Atoms, Composites, and Conduits—so boundaries, dependencies, and integration points stay explicit as the product grows. The skill guides architecture design sessions and produces documentation that scales with maturity: lighter sketches early, richer ADRs and composite maps when stakes are higher. It fits builders shipping SaaS, APIs, or agent-heavy systems who need a repeatable way to capture why structures were chosen and how components compose. Use it when a feature set is taking shape and you want decisions durable enough for future you, collaborators, or coding agents to follow without re-deriving the system from chat history.0installs3Exeexe is a Lucid toolkit workflow plugin for solo builders who already have a tech spec and need a dependable way to break it into executable phases. Instead of ad-hoc prompting, it produces validated execution plans, then drives the agent through those phases with parallel task groups so independent work can move at once. Progress and decisions accumulate in execution logs, which makes it easier to resume after interruptions, compare iterations, and hand off context to review or ship steps. It is aimed at indie developers shipping SaaS, CLIs, or agent-heavy projects where scope is defined but the order of operations still matters. Use it when you want structured runbooks tied to specs—not one-shot code dumps—and when you care about traceability from plan to done.0installs4Impl Flutterimpl-flutter is a large Lucid-toolkit plugin for solo builders shipping cross-platform mobile apps who need more than a single “build my screen” prompt. Twenty-six specialized agents split planning, execution, codegen, generative UI, testing, debugging, verification, persistence, platform hooks, UX polish, releases, and environment setup—so you can treat Flutter work like a staged pipeline instead of improvising every session. Canonical placement is Build / frontend because Dart widgets, Riverpod state, and feature delivery dominate, but planning orchestration aligns with Validate-style scoping and E2E or release agents naturally extend into Ship. It suits indie founders building consumer or B2B mobile clients, especially when you already chose Flutter and want TDD-leaning discipline. It is not a substitute for market validation or store-launch playbooks, but it gives repeatable agent choreography from plan to verifiable mobile increments.0installs5Impl Neo4jimpl-neo4j is an agent skill packaged as a Neo4j specialist plugin that delegates hard graph work across six focused agents—modeling, Cypher authoring, data movement, performance tuning, driver wiring, and environment setup. Solo builders reach for it when a relational mindset breaks on relationship-heavy domains like recommendations, permissions graphs, or knowledge graphs. Instead of one generic database prompt, you get procedural depth on Cypher patterns, import/export mechanics, and tuning queries before they stall production. The toolkit tags signal APOC, GDS, and MCP usage, so the skill fits agent-led builds where the database is queried and evolved iteratively. Use it while implementing backend features that persist and traverse connected data, not as a one-off SQL migration helper.0installs6Impl Pythonimpl-python is a Lucid-toolkit implementation plugin aimed at solo and indie builders who want structured Python delivery without juggling unrelated prompts. It bundles eight specialized agents that mirror how a small team would divide work: generate code, run tests, debug failures, wire SQLAlchemy persistence, integrate platforms, shape APIs, cut releases, and stabilize environments. The skill assumes you are past vague ideation and actively building backends or API-backed products where FastAPI and Pydantic are reasonable defaults. Use it when you need consistent implementation rituals rather than one-off chat snippets—especially for services, internal tools, and SaaS backends. It does not replace product validation or security review, but it compresses the Build phase into agent-sized steps that map cleanly to backend and integration subphases, with natural spillover into Ship when testing and release agents run.0installs7LucLuc is a Claude Code plugin skill from the lucid-toolkit that bundles status line, schema-based output, TOON formatting, and workspace utilities for solo builders who want their coding agent to stay oriented and produce consistent structured results. Install it when you are standardizing how Claude Code reports state, validates payloads, or navigates a monorepo-style workspace—not when you need a one-off app feature. It fits the Build phase as agent-tooling and also helps during Ship and Operate when you want repeatable setup and readable agent transcripts. Intermediate complexity: you should already be comfortable with Claude Code plugins and local workspace conventions. Pair it with other agent skills once Luc is configured so downstream steps inherit the same output discipline.0installs