Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/jasonkneen/openclicky/llms.txt

Use this file to discover all available pages before exploring further.

OpenClicky ships over 50 bundled skills covering the most common developer and productivity workflows. Every skill is a SKILL.md file maintained in the app bundle — ready to use without any setup beyond the prerequisites each skill documents. This page is a complete reference organized by category.
Skill descriptions are taken directly from each skill’s SKILL.md frontmatter. The name shown is the machine identifier used in related_skills references.

Coding & Development

These skills delegate coding tasks to autonomous coding agents, manage Git workflows, and help with code review. They coordinate through the OpenClicky terminal and, where needed, via tmux sessions for interactive multi-turn work.
SkillNameDescription
claude-codeclaude-codeDelegate coding to Claude Code CLI (features, PRs). Supports print mode for one-shot tasks and interactive PTY mode via tmux for multi-turn sessions.
codexcodexDelegate coding to OpenAI Codex CLI (features, PRs).
github-authgithub-authGitHub auth setup: HTTPS tokens, SSH keys, gh CLI login.
github-code-reviewgithub-code-reviewReview PRs: diffs, inline comments via gh or REST.
github-issuesgithub-issuesCreate, triage, label, assign GitHub issues via gh or REST.
github-pr-workflowgithub-pr-workflowGitHub PR lifecycle: branch, commit, open, CI, merge.
github-repo-managementgithub-repo-managementClone/create/fork repos; manage remotes, releases.
openclicky-repo-operatoropenclicky-repo-operatorOperate on software repositories and GitHub work: clone/open repos, inspect codebases, explain structure, create branches, commit, push, open PRs, review diffs, and debug CI failures.
openclicky-dev-setup-doctoropenclicky-dev-setup-doctorDiagnose and fix developer environment, agent runtime, MCP, API key, package manager, localhost, Node/npm/Python, Supabase, Cloudflare/Wrangler, Codex, Claude Code, and terminal setup problems.
openclicky-build-previewopenclicky-build-previewBuild, modify, launch, preview, and iterate websites, web apps, dashboards, landing pages, HTML files, React/Next apps, and frontend UI. Use when the user wants a visible working thing, not only code.
claude-code is OpenClicky’s primary autonomous coding skill. It delegates work to Claude Code — Anthropic’s coding agent CLI — and provides two orchestration modes:
  • Print mode (-p) — preferred for one-shot tasks: fix a bug, add a feature, review a diff. No PTY needed, no interactive dialogs.
  • Interactive PTY via tmux — for multi-turn iterative sessions: refactor → review → test cycles where you need slash commands or mid-session follow-up.
The skill covers PTY dialog handling, --allowedTools scoping, session resumption with --resume, structured JSON output, parallel Claude instances, CLAUDE.md project memory, custom subagents, hooks, and MCP server integration.Related: codex, opencode
github-issues handles the full GitHub Issues lifecycle. Every operation is shown with two paths:
  1. gh CLI — preferred when authenticated via gh auth login
  2. curl + REST API — fallback using $GITHUB_TOKEN
Operations covered: listing and filtering issues, creating issues with bug-report and feature-request templates, adding/removing labels, assigning issues, commenting, closing and reopening, linking issues to PRs with closing keywords, creating branches from issues with gh issue develop, and bulk operations.Related: github-auth, github-pr-workflow

Screen & UI Automation

These skills control macOS apps and browser UIs in the background using OpenClicky’s Computer Use runtime — without stealing focus from whatever you’re currently working on.
SkillNameDescription
cua-drivercua-driverDrive native macOS apps and browser UI through OpenClicky’s local computer-use MCP server, backed by Cua. Snapshot AX state, act by element index, keep work backgrounded, prefer the user’s default browser, and verify via re-snapshot.
openclicky-screen-controlopenclicky-screen-controlInstantly control OpenClicky’s local overlay bridge to point on screen, show captions, or speak through OpenClicky’s voice. Use when the user says “show me where you mean”, “point to it”, “highlight that”, or asks for visual on-screen guidance.
openclicky-screen-touropenclicky-screen-tourCreate recordable OpenClicky visual tours with multiple simultaneous markers, area-focused overlays, primary cursor choreography, screenshots, captions, and TTS.
animateanimateReview a feature and enhance it with purposeful animations, micro-interactions, and motion effects that improve usability and delight. Use when the user mentions adding animation, transitions, micro-interactions, motion design, hover effects, or making the UI feel more alive.
cua-driver is OpenClicky’s core Computer Use skill. Its central rule is the no-foreground contract: the user’s frontmost app must never change while the agent operates another app. This is enforced by:
  • Using launch_app({bundle_id}) instead of any form of open — which always activates and foregrounds
  • Using element_index-based AX actions instead of pixel coordinates
  • Calling get_window_state(pid, window_id) before AND after every action to verify state
The skill maps common intents to their correct Computer Use MCP tools:
IntentCorrect toolForbidden
Launch an applaunch_app({bundle_id})open -a, osascript activate
Click a UI elementclick({pid, window_id, element_index})cliclick, coordinate clicks
Type texttype_text({pid, text})osascript keystroke
Navigate a browser URLlaunch_app({bundle_id, urls: [...]})⌘L, set_value on omnibox
openclicky-screen-control talks to a local HTTP bridge at http://127.0.0.1:32123. Key endpoints:
  • POST /cursor — point OpenClicky’s triangle cursor at a screen coordinate with an optional caption
  • POST /cursors — show multiple simultaneous cursors for comparing alternatives
  • POST /caption — show a floating text caption near a coordinate
  • POST /screenshot — capture the screen and return AppKit coordinate metadata
  • POST /speak — speak text through OpenClicky’s voice without entering voice mode
  • POST /clear — clear all overlay elements
Coordinates use macOS/AppKit space: origin at the bottom-left of the global desktop.

Productivity & Docs

Skills for creating, reading, converting, and saving documents in common formats, plus personal knowledge management.
SkillNameDescription
docdocCreate and edit DOCX documents.
pdfpdfUse when tasks involve reading, creating, or reviewing PDF files where rendering and layout matter. Uses Poppler for visual checks and Python tools (reportlab, pdfplumber, pypdf) for generation and extraction.
spreadsheetspreadsheetUse when tasks involve creating, editing, analyzing, or formatting spreadsheets (.xlsx, .csv, .tsv) with formula-aware workflows, cached recalculation, and visual review.
powerpointpowerpointCreate, read, edit .pptx decks, slides, notes, templates.
ocr-and-documentsocr-and-documentsExtract text from PDFs/scans using pymupdf and marker-pdf.
savesaveIngest links, screenshots, and notes into a personal knowledge wiki. Use when the user says “save”, “remember”, “note this”, or wants to capture information for later.
save-wikisave-wikiWrite counterpart to read-wiki. Ingest links, screenshots, and notes into the personal knowledge wiki — a persistent, compounding artifact inspired by Karpathy’s LLM Wiki pattern.
read-wikiread-wikiUse when a task references a specific project, person, or note the user has saved. Read-only counterpart to save-wiki.
openclicky-research-reportopenclicky-research-reportResearch a topic, market, competitor set, product, person, or question and turn findings into a Markdown, PDF, DOCX, CSV, or brief artifact.
openclicky-email-assistantopenclicky-email-assistantDraft, rewrite, summarize, triage, and prepare replies or outreach emails. Use for Gmail/Outlook/Mail tasks, email thread summaries, outbound sequences, follow-ups, and humanizing drafts.

Google Workspace

Two complementary skills for Google Workspace access, both built on the local gog CLI from gogcli.
SkillNameDescription
google-workspace-gogcligoogle-workspace-gogcliConnect to Google Workspace through the local gog command for Gmail, Calendar, Drive, Docs, Sheets, Slides, Chat, Contacts, Tasks, Groups, Admin, Classroom, Forms, Apps Script, People, and Keep. Full setup and auth documentation included.
goggogUse OpenClicky’s local gog CLI as the primary Google Workspace route for Gmail read/search, Calendar, Drive, Docs, Sheets, Chat, Contacts, auth status, and account inspection. Prefer this over browser automation for normal Google Workspace work.
Both skills use gog — a local CLI that stores credentials in the macOS Keychain (or an encrypted file keyring). Auth is handled locally; no credentials are stored in the OpenClicky cloud.To set up:
# Check if gog is installed
command -v gog && gog --version

# Install if missing (macOS)
brew install gogcli

# Check auth status
gog --json auth status

# Add an account with least-privilege scopes
gog auth add you@example.com --services gmail,calendar,drive --gmail-scope readonly
For normal Gmail/Calendar/Drive tasks, auth belongs in OpenClicky Settings → Google. The skills only provide manual gog auth commands when the user explicitly asks for setup help.

Third-Party Apps

Skills that connect Clicky to popular productivity and communication apps via their APIs or local CLIs.
SkillNameDescription
notionnotionNotion API via curl: pages, databases, blocks, search. Requires NOTION_API_KEY.
linearlinearLinear: manage issues, projects, teams via GraphQL + curl. Requires LINEAR_API_KEY.
airtableairtableAirtable REST API via curl. Records CRUD, filters, upserts. Requires AIRTABLE_API_KEY.
obsidianobsidianRead, search, create, and edit notes in the configured Obsidian vault via filesystem-first access.
apple-notesapple-notesManage Apple Notes via memo CLI: create, search, edit. macOS only.
apple-remindersapple-remindersApple Reminders via remindctl: add, list, complete. macOS only.
imessageimessageSend and receive iMessages/SMS via the imsg CLI on macOS.
findmyfindmyTrack Apple devices/AirTags via FindMy.app on macOS.
mapsmapsGeocode, POIs, routes, timezones via OpenStreetMap/OSRM.
spotifyspotifySpotify: play, search, queue, manage playlists and devices.
polymarketpolymarketQuery Polymarket: markets, prices, orderbooks, history.
The notion skill targets Notion API version 2025-09-03. In this version, databases are called data sources in the API — use /data_sources/ endpoints for queries, but still use database_id when creating pages. Each database carries both a database_id and a data_source_id; they’re used in different contexts.

AI & Creative

Skills for UI design, diagramming, 3D work, and creative content generation.
SkillNameDescription
hallmarkhallmarkAnti-AI-slop design skill for greenfield pages, audits, redesigns, and design extraction from URLs or screenshots. Use when the user asks to build a new app or landing page, wants to redesign something, invokes Hallmark by name, or uses audit/redesign/study.
frontend-designfrontend-designBuild or improve frontend UI for websites, apps, dashboards, landing pages, and local previews. Use when the user asks for a working interface, visual redesign, UI polish, layout fixes, responsive behavior, or purposeful animation.
claude-designclaude-designDesign-focused skill for UI/UX work using Claude.
excalidrawexcalidrawHand-drawn Excalidraw JSON diagrams (architecture, flow, sequence).
blenderblender-toolkitBlender automation with geometry creation, materials, modifiers, and Mixamo animation retargeting. WebSocket-based real-time control, automatic bone mapping, two-phase confirmation workflow.
hallmarkhallmarkAnti-AI-slop design skill with 22 named themes, structural macrostructures, pre-emit slop tests, and a strict no-fabricated-content rule.
hatch-pethatch-petCreate, repair, validate, preview, and package Codex-compatible animated pet spritesheets from character art, screenshots, or visual references.
openclicky-creative-studioopenclicky-creative-studioRoute broad creative work to available OpenClicky capabilities: frontend polish, design critique, brand/social planning, document/PDF outputs, spreadsheet/table outputs, and existing visual artifact handling.
openclicky-artifactsopenclicky-artifactsOpen, reveal, find, export, rename, move, organize, or explain existing OpenClicky-generated files and local artifacts — PDFs, CSVs, DOCXs, PPTXs, images, reports.
animateanimateEnhance features with purposeful animations and micro-interactions that improve usability and delight.
polishpolishPerforms a final quality pass fixing alignment, spacing, consistency, and micro-detail issues before shipping.
Hallmark ships four verbs on top of its default “build something new” behavior:
InvocationEffect
(default)Full design flow: pre-flight scan → audience/use-case/tone → macrostructure → theme → build → 69-gate slop test
hallmark audit <target>Score the target against the anti-pattern list; return a ranked punch list. No edits.
hallmark redesign <target>Redesign the visual structure within existing implementation boundaries. Preserves routes, component ownership, and copy intent.
hallmark study <screenshot|URL>Extract the design DNA — macrostructure, archetypes, type-pairing, colour anchor — from a reference. Optionally rebuild using the extracted DNA.
Hallmark enforces rules like: no invented metrics, no re-drawn browser chrome, locked OKLCH tokens, mandatory 8-state interactive components, and mobile breakpoints verified at 320/375/414/768 px.

OpenClicky System

Skills that manage OpenClicky itself — tutorials, self-improvement, onboarding, and agent management.
SkillNameDescription
openclicky-guided-tutorialsopenclicky-guided-tutorialsUse OpenClicky’s pointing tool calls, screenshots, captions, and speech to walk the user through visible macOS or app workflows step by step.
openclicky-specialist-agentsopenclicky-specialist-agentsCreate, equip, or repair OpenClicky specialist agents with explicit soul, instructions, memory, heartbeat, and skills.json wiring.
learn-from-openclicky-logslearn_from_openclicky_logsUse when the user asks OpenClicky to review logs, find learnings, tune behavior from logs, or create improvements from logged messages.
optimize-openclicky-skillsoptimize_openclicky_skillsUse when the user asks OpenClicky to inspect, audit, improve, consolidate, or optimize bundled or learned skills.
create-onboarding-hello-worldcreate-onboarding-hello-worldBuild the small “hello ” website that OpenClicky can create as a quick starter task.
vercel-deployvercel-deployDeploy applications and websites to Vercel. Always deploys as preview by default unless the user explicitly asks for production.
youtube-contentyoutube-contentYouTube transcripts to summaries, threads, and blog posts. Use when the user shares a YouTube URL or asks to summarize a video.
openclicky-email-assistantopenclicky-email-assistantDraft, rewrite, summarize, triage, and prepare replies or outreach emails with explicit approval before sending.
openclicky-research-reportopenclicky-research-reportResearch a topic and produce a Markdown, PDF, DOCX, CSV, or brief artifact using search, fetch, and local extraction.
These two system skills form OpenClicky’s self-improvement loop:
  1. learn-from-openclicky-logs — reads your conversation history (JSONL log files) and extracts repeatable patterns. For each identified pattern, it can draft a new SKILL.md in your OpenClickyLearnedSkills/ directory.
  2. optimize-openclicky-skills — reviews existing skills (both bundled and learned) and improves them based on evidence from usage. The safe protocol is: archive the old SKILL.md first, then overwrite — this preserves the original in case the optimization degrades performance.
To trigger either skill, tell Clicky naturally:
  • “Review my logs and create learnings from them.”
  • “Optimize my skills based on how they’ve been performing.”

Build docs developers (and LLMs) love