dotfiles

Summarize the current conversation (or part of it) in a publishable markdown file in the agent voice for the user to publish verbatim. DO NOT use this skill to ghostwrite in the user's voice or to draft original blog posts - it is strictly for summarizing a chat that already happened. Use when user asks for a write up or summary of the chat.

Collaborate with the user asynchronously through a shared markdown file in the working tree. The agent writes questions or proposals in the file and stops; the human replies at their own pace, freeform. Use when the user mentions "lazychat" or asks to "discuss in a file", when the task has several entangled decisions better thought through than chatted through, or when the user may want to step away and reply later. Do NOT use for single-question decisions, bug fixes, or when the user is actively present and driving. Requires the lazychat CLI to be installed.

Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me".

Rewrite documents so AI agents follow them more reliably. Use when optimizing CLAUDE.md, AGENTS.md, READMEs, runbooks, specs, or any instructional document for AI consumption. Applies context engineering patterns: attention-aware structure, positive framing, token efficiency, canonical examples, consistent terminology, and explicit degrees of freedom. Do NOT use on code files, configs, or content that humans read but agents do not.

Extract and distill project knowledge for Claude.ai Projects. Use when the user wants to generate or update a set of files to upload to a Claude.ai Project so they can discuss the codebase in claude.ai conversations. Triggers on requests like "sync to claude project", "generate project knowledge", "update claude.ai project files", or "extract knowledge for claude.ai".

Generate multiple radically different interface designs for a module using parallel sub-agents. Use when user wants to design an API, explore interface options, compare module shapes, or mentions "design it twice".

Diátaxis Documentation Expert. An expert technical writer specializing in creating high-quality software documentation, guided by the principles and structure of the Diátaxis technical documentation authoring framework.

Verify factual claims in text before delivering it to the user. Use this skill whenever you are writing, reviewing, or editing blog posts, technical specs, tool comparisons, documentation, or any content that makes verifiable claims. Also trigger when the user mentions fact-checking, verification, AI slop, hallucination detection, or asks you to review content for accuracy. This skill should run as a final pass before presenting written content. Do not skip it just because you feel confident about the claims.

Flush working context to disk at any point during a session. Use when the user says flush, checkpoint, save progress, save context, let's wrap up, end of session, or signals a pause. Also invoke proactively when context is getting long and the user hasn't explicitly saved progress.

Code review for Go CLI applications. Produces an actionable checklist covering error handling, CLI framework patterns (Cobra/urfave), testing, performance, security, and Go idioms. TRIGGER when: reviewing Go CLI code, auditing a Cobra/urfave app, or asked to check a Go command-line tool. Do NOT TRIGGER when: writing new Go code (use golang-pro), reviewing non-CLI Go projects, or general code review without a CLI focus.

Use when building Go applications requiring concurrent programming, microservices architecture, or high-performance systems. Invoke for goroutines, channels, Go generics, gRPC integration.

Remove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's Signs of AI writing guide. Detects and fixes patterns including: inflated symbolism, promotional language, superficial -ing analyses, vague attributions, em dash overuse, rule of three, AI vocabulary words, negative parallelisms, and excessive conjunctive phrases. Do NOT use on technical documentation, API docs, or code comments. Only use on prose intended for human readers.

Automate pull request creation with intelligent verification and semi-interactive workflow. Use when ready to open a PR after completing work. The skill verifies task completion from conversation history, checks unpushed code, reads the repo's PR template, generates a suggested title and description combining git log and conversation context, and opens a draft PR via gh. Semi-interactive shows the proposal for confirmation/editing before creating.

Structured project inception through interview, documentation, and scaffolding. Use when starting a new project (CLI tools, libraries, web apps, backends, etc.) to conduct a detailed interview about vision/scope/tech, generate foundational documentation (README with philosophy, CLAUDE.md for AI context, ROADMAP for phased plan), and scaffold initial code structure with git setup. Adapts questions and outputs based on project type. Do NOT use for existing projects. Use only when starting from scratch.

Disciplined implementation workflow following the red-green-refactor-commit cycle. Invoke explicitly via /red-green-refactor or suggest when the user describes a non-trivial multi-file implementation task, feature addition, or significant code change. Do NOT suggest for single-file fixes, typos, or trivial changes.

Run a security and health audit on a GitHub repository using OpenSSF Scorecard. Analyzes results in context - a weekend CLI tool gets different priorities than a production service. Use this skill whenever the user mentions repo security, branch protection, security audit, scorecard, supply chain security, repo hardening, dependency pinning, or wants to check if their GitHub project follows security best practices. Also trigger when the user asks "is my repo secure?" or "what should I fix first?" about a GitHub repository.

Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.

Create technical specifications for features and refactoring. Use when designing new features, refactoring components, or planning architectural changes. Guides full spec creation from concept to implementation with test-first approach, proper formatting, task lists, and dependency tracking. Do NOT use for bug fixes, small changes, or tasks that fit in a single session without planning.

Plan and execute risky technical changes by identifying the riskiest assumption, designing a minimal end-to-end test that validates it, then growing the implementation in phased red-green-refactor cycles. Produces a persistent plan file for multi-session continuity. Use when: (1) a proposed approach has an assumption that could invalidate the entire design, (2) migrating to a new architecture or integration pattern, (3) adopting an unfamiliar technology or API, (4) any multi-session implementation where validating feasibility early saves significant rework. Trigger phrases: "tracer bullet", "validate the approach", "riskiest assumption", "prove this works first", "thin vertical slice". Do NOT use for: single-file changes, well-understood patterns, tasks where the approach is known to work, or pure refactoring of existing code.

Extract a DDD-style ubiquitous language glossary from the current conversation, flagging ambiguities and proposing canonical terms. Saves to UBIQUITOUS_LANGUAGE.md. Use when user wants to define domain terms, build a glossary, harden terminology, create a ubiquitous language, or mentions "domain model" or "DDD".