agentics

Marks a plan as completed. Verifies codebase evidence, ticks acceptance criteria, and re-renders the HTML; --all sweeps unmarked plans. Use via /plan-agent:finalize-plan.

Generates HTML implementation-plan documents. Produces a self-contained .html plan file with steps, acceptance criteria, and metadata. Use when the user asks to create or generate an HTML plan file.

Stress-tests plan decisions node-by-node with focused questions. Walks each decision point surfacing assumptions and weak spots. Use when the user asks to deep grill or stress-test a plan.

Generates a prose reference doc from a completed plan. Inspects the codebase and git history to produce accurate, evidence-backed documentation. Use when the user asks to document a completed plan.

Converts a markdown file into a self-contained HTML page. Produces styled output with syntax highlighting and responsive layout. Use when the user asks to convert a markdown file or plan to HTML.

Writes lifecycle status into a plan's frontmatter, one file or a directory. Inspects codebase and git history for accurate dates. Use when asked to check or update plan status.

Plan review Agent Team. Reviews HTML implementation plans in parallel, synthesizes findings, and applies improvements in place. Use when the user asks to review or improve an implementation plan.

Runs a six-role Agent Team to review product plans in place. Covers PM, Dev, UX, Frontend, A11y, and Security on PRDs and proposals. Use when the user asks to review or improve a product plan.

Writes developer guides on any project topic. Covers rules, tools, resources, plans, and changes for any reader. Use when the user asks to write, explain, or deep-dive a project topic as a guide.

Installs the team's shared rule files into ~/.claude/rules/. Copies the bundled plan-mode, UI, JSDoc, and review rules with per-file confirmation. Use when asked to sync team rules.

Runs the full ship pipeline with verification, CI polling, and bounded autofix. Chains tests, preview, commit, PR, CI poll, and gated merge. Use when asked to autonomously ship or watch CI.

Turns a vague idea into a decision-complete proposal. Researches web and codebase, separating established facts from open decisions. Use when the user floats an idea or asks should-we.

Generates a runnable static-HTML prototype from a plan, idea, or design. Produces one self-contained, framework-free clickable file under docs/prototypes/. Use when asked to prototype a mockup.

Exports a Claude Code session transcript to Markdown. Converts session JSONL into readable reference material in the plans directory. Use when asked to export or archive a session.

Saves an HTML Artifact page to the local artifacts inbox. Copies a chosen .html into .claude/artifacts with a dated name, then publishes it. Use when asked to save or share an artifact.

Generates social copy and a dark-mode card image for code changes. Detects changes from git, picks a template, and screenshots via Playwright. Use when asked to share a code change.

Publishes a session recap as a claude.ai artifact. Extracts transcript turns into a Summary, Decisions, and Learnings recap, scrubs it, then publishes. Use when asked to share a session recap.

Publishes saved prompts as claude.ai artifacts. Copies raw prompt text verbatim; publishes one prompt or the whole filterable library. Use when asked to publish or share a prompt.

Publishes an annotated diff walkthrough as a claude.ai artifact. Scrubs for secrets, then builds a self-contained page with per-hunk reviewer notes. Use when asked to publish or share a diff.

Publishes an implementation plan as a claude.ai artifact. Republishes to the same URL across sessions so viewers watch steps check off live. Use when asked to publish or share a plan.

Builds structured AI prompts using Anthropic techniques. Interviews users, classifies prompt type, and delivers a copy-pasteable prompt. Use when the user asks to write, refine, or build a prompt.

Creates a git branch from origin/<default> with no upstream tracking. Auto-names the branch from staged or unstaged changes in the working tree. Use when the user asks to create or start a new branch.

Stages all changes and creates a conventional commit message. Analyzes the diff and writes a descriptive, scope-correct commit. Use when the user asks to commit or save work to git.

Drafts and opens a GitHub or GitLab issue from any context source. Detects host from git remote and confirms before creating. Use when the user asks to file, open, or create an issue or ticket.

Pushes the branch and creates a pull request. Supports GitHub and GitLab via gh and glab with auto-filled title and body. Use when the user asks to create a PR or open a pull request.

Builds and opens a filterable HTML gallery of all plans in the plans directory. Scans HTML plans, parses metadata, and writes index.html. Use when asked to browse plans or view plan history.

Builds and opens a filterable HTML gallery of saved social cards. Scans docs/media/social/ and generates a filterable index page. Use when asked to browse the media library or view saved posts.

Opens the existing plans gallery without rebuilding. Opens index.html directly; run plans-library first if not yet built. Use when asked to reopen the plans gallery or browse plans without a rebuild.

Scaffolds GitHub Pages publishing into any repo. Adds the deploy workflow, .nojekyll, hub, and preview script so docs/ HTML reaches a public URL. Use when asked to set up or publish GitHub Pages.

Explains how any project file, component, or concept works. Reads source files, synthesizes principles, and generates a social card. Use when the user asks how something works or to explain it.

Turns a React component into a social card with preview, code, and props table. Builds a static preview and screenshots react-card.html via Playwright. Use when asked to share a React component.

Social media share router — classifies content type and runs the right skill. Dispatches to the right share-* skill based on content type. Use when asked to share what you're working on.

Implements features via TDD with up to 20 red-green-refactor rounds. Writes failing tests first, then iterates until all pass. Use when the user asks to TDD a new feature or write tests first.

Ships changes by staging, committing, pushing, and opening a PR. Supports GitHub and GitLab in a single guided flow. Use when the user asks to ship changes or commit and create a PR.

Optimizes SKILL.md frontmatter fields. Rewrites descriptions to three-part format (≤200 chars) and tunes disable-model-invocation. Use when the user asks to optimize SKILL.md frontmatter.

Creates platform-aware social copy and a dark-mode card for a blog post. Fetches OG tags, populates blog-card.html, and screenshots via Playwright. Use when asked to share a blog post on social media.

Fetches a GitHub file and generates a syntax-highlighted social card with copy. Scrubs, fills snippet-card.html, and screenshots via Playwright. Use when asked to share a code snippet from GitHub.

Generates a SOCIAL.md project sharing config by analyzing the codebase. Use when asked to set up social sharing preferences or create a SOCIAL.md file.

Generates social posts and dark-mode cards for project updates. Reads git, CHANGELOG, and README then screenshots via Playwright. Use when announcing features, bugs, or releases on social media.

Scans git history or a codebase path for shareable code. Drafts social media prompts for share-code or digest generation. Use when the user asks to find commits worth sharing or create a digest.