claude-skills
Agent skills by @shaharsha - 24 production-grade Agent Skills that work in Claude Code, claude.ai, Codex, Cursor, and any other harness that reads the SKILL.md format.
MIT licensed. Built day-to-day; battle-tested in real projects.
Install
Claude Code (recommended)
/plugin marketplace add shaharsha/claude-skills
/plugin install shaharsha-skills@shaharsha-skills
This installs all 24 skills as one plugin. Or pick a subset:
/plugin install documents-and-decks@shaharsha-skills # gdoc-sync + gslides-sync + gsheets + presentation-generator + elevenlabs-tts + narrating-pptx + deck-to-video + self-presenting-decks
/plugin install brand-and-visuals@shaharsha-skills # brand-system + brand-assets + image-generation + excalidraw-diagrams
/plugin install engineering-decisions@shaharsha-skills # tech-design-doc + codex-review
/plugin install building-agents@shaharsha-skills # prompt-engineer + writing-project-instructions + reviving-sessions
/plugin install utilities@shaharsha-skills # namecheap-domains + office-render + viewing-videos + proton-pass + tavily-extract
/plugin install video-production@shaharsha-skills # generating-video-clips + producing-animated-episodes
Manual (any other harness)
git clone https://github.com/shaharsha/claude-skills.git ~/shaharsha-skills
ln -s ~/shaharsha-skills/skills/<skill-name> ~/.claude/skills/<skill-name>
# repeat per skill, or symlink the whole skills/ dir
For Codex, Cursor, Gemini CLI, OpenClaw, etc., follow the same shape - point your harness at skills/<name>/SKILL.md.
Skills
Each links to its own README — what it does, why it exists, install, and the gotchas it encodes.
Documents & decks
| Skill | What it does |
|---|
| gdoc-sync | Push a local .md to an existing Google Doc, fixing the four things Google's converter breaks: #anchor links, cross-doc refs, 1500pt images, and missing RTL. |
| gslides-sync | The same shape for .pptx → an existing Google Slides: native pageObjectId links, scaled images, RTL per text shape. |
| gsheets | Full Sheets API v4 CLI — cells, tabs, freeze/resize/merge, header styling, filters, banding, conditional formatting, plus a raw batchUpdate escape hatch. |
| presentation-generator | 16:9 PDF + PPTX decks where every slide is a custom AI-rendered image. Style locks globally; composition varies per slide. |
| elevenlabs-tts | Speech with ElevenLabs Eleven v4 in any language — v4 audio-tag direction, consistent 429-safe batches, a speech-to-text check that no tag was read aloud, and forced-alignment timings for subtitles. |
| narrating-pptx | Per-slide ElevenLabs narration embedded in a pptx, with autoplay authored by real PowerPoint — the one method that doesn't corrupt the file. |
| deck-to-video | Slides + narration → a self-playing mp4 with a progress bar, countdown, and slide counter, baked with PIL because ffmpeg's animated drawbox silently renders full. |
| self-presenting-decks |
Brand & visuals
| Skill | What it does |
|---|
| brand-system | A production-grade brand book: 20-section BRAND.md, printable PDF sibling, and matched tokens.css (Tailwind v4) + tokens.json (W3C DTCG). WCAG 2.2 AA audited at authoring time. |
| brand-assets | The mechanical-pixel sibling: vectorize, finalize-svg, rasterize, icon-pack, colour-audit. Bash + Python stdlib. |
| image-generation | Logos, icons, mockups, and product shots via OpenAI GPT Image 2.5 (Sunburst for finals, Flare for drafts) or Gemini Nano Banana — each prompted in its provider's native grammar, with native transparent PNGs. |
| excalidraw-diagrams | Architecture and flow diagrams on an Excalidraw+ canvas with real tech icons, plus the MCP quirks that silently break a diagram. |
Engineering decisions
| Skill | What it does |
|---|
| tech-design-doc | RFCs, ADRs, KEPs, and partner-mode TDRs, triaged to the right format first and audited against documented anti-patterns. |
| codex-review | An independent read-only second opinion on a plan or diff from OpenAI Codex — then adjudicate every finding against the source. |
Building agents
| Skill | What it does |
|---|
| prompt-engineer | Prompt and context engineering for the Claude / GPT / Gemini / Kimi / Grok APIs — system prompts, tool descriptions, provider differences, judge-prompt design. |
| writing-project-instructions | Author or audit a CLAUDE.md / AGENTS.md, with the falsifiability test and the loading semantics people usually get wrong. |
| reviving-sessions | Bring back sessions the Claude desktop app cannot see — adopt CLI and headless sessions into it, fix renames, recover stranded transcripts, and read what an ended session said. |
Utilities
| Skill | What it does |
|---|
| namecheap-domains | Domain availability via Namecheap's API — one domain, batches of 50, or TLD sweeps, with premium and EAP fees surfaced. |
| office-render | Render .docx / .pptx / .xlsx to PDF and page images using the real Office apps on macOS — pixel-faithful, unlike LibreOffice. |
| viewing-videos | How an agent sees a video: the fewest ffmpeg frames that answer the question, triaged on a contact sheet. |
| proton-pass | Credentials and secrets from the CLI via Proton Pass, with never-print-secrets guardrails. |
| tavily-extract | Fetch the JS-rendered SPAs and 403-blocked hosts a normal fetcher can't, and flag soft 404s. |
Video production
| Skill | What it does |
|---|
| generating-video-clips | Still to short clip with Gemini Omni, takes reviewed on frame sheets, and mouths lip-synced to your own dialogue with Sync 3 on fal, one speaker per part, frame-exact. |
| producing-animated-episodes | A voiced, subtitled, animated, lip-synced vertical episode from photos and a script: one episode.json, one engine, human gates at every paid step, timing driven by the real dialogue. |
How they compose
A few of these are designed to work together:
presentation-generator calls image-generation for every slide.presentation-generator consumes brand-system's BRAND.md for palette / typography / motif lock when present in the directory.brand-system (the document) and brand-assets (the pixels) are siblings - run both for a complete brand rollout.gdoc-sync, gslides-sync, and gsheets share Google service-account setup; one SA works for all three APIs.tech-design-doc calls gdoc-sync at the end of the workflow to push the finished TDR to a live Google Doc for stakeholder comments.narrating-pptx narrates decks produced by presentation-generator (or any pptx), generates its audio with elevenlabs-tts, and uses office-render for the real-PowerPoint validation step.deck-to-video reuses the per-slide mp3s (and elevenlabs-tts' forced alignment for subtitles) and office-render's real-PowerPoint PDF — the pptx and the video sound and look identical.self-presenting-decks is the map over the whole chain: pptx authoring → office-render → narrating-pptx (+ elevenlabs-tts) → deck-to-video.generating-video-clips animates image-generation's panels and lip-syncs them to elevenlabs-tts dialogue.producing-animated-episodes drives image-generation (cards, panels), elevenlabs-tts (voices, music, SFX) and generating-video-clips (clips, lip-sync) from one config.
Compatibility
These are standard Agent Skills (SKILL.md-format). They work in any harness that reads the format:
- Claude Code (CLI) - full plugin marketplace support; install via
/plugin install as above.
- claude.ai - upload skills via Settings -> Capabilities -> Skills.
- Codex CLI - point Codex at the skill folder; it reads SKILL.md natively.
- Cursor - drop the skill into your
~/.cursor/skills/ (or wherever your Cursor config expects).
- Gemini CLI - point at the skill via
gemini extensions install.
- OpenClaw / other harnesses - any tool that respects SKILL.md frontmatter works.
License
MIT. See LICENSE.
