Claude Code skills for developer productivity, research, communication, and workflow automation.
claude plugin marketplace add aliir74/cortexclaude plugin install cortexAfter this, updates pull automatically at Claude Code startup.
Invoke skills with /cortex:<skill-name>.
| Skill | Description | Usage |
|---|---|---|
babysit-pr |
Monitor a PR — auto-fix CI failures, address review feedback, track deploys | /loop 5m /cortex:babysit-pr #123 |
codex-ask |
Get a second opinion from OpenAI Codex CLI | /cortex:codex-ask Is this approach correct? |
commit-push-pr |
Stage, commit, push, and open a PR/MR in one flow (auto-detects GitHub vs GitLab) | /cortex:commit-push-pr |
helpers |
Shared stdlib-only scripts (adherence probes, corpus measurement, skill audit, safe description edits) used by other skills; not a workflow on its own | python3 <plugin>/skills/helpers/<script> --help |
make-slides |
Build an HTML slide deck from the MIT beautiful-html-templates library (by Zara Zhang) | /cortex:make-slides pitch deck for X |
name-session |
Suggest a structured [context] goal session name and copy the /rename command |
/cortex:name-session |
new-session |
Write a task or continuation brief and launch a new background Claude Code session on it in another folder | /cortex:new-session ~/code/other-repo |
session-handoff |
Generate structured handoff document for another agent/engineer | /cortex:session-handoff |
These skills activate automatically when Claude detects you're working in a relevant area:
| Skill | Triggers When |
|---|---|
bird-cli |
Interacting with Twitter/X — reading, searching, posting, replies, bookmarks |
catch-up |
Re-briefing you on the current (resumed or long) session — "catch me up", "where were we" |
clickup-cli |
Running ClickUp operations — tasks, comments, search, sprints, time tracking |
compare-skillsets |
Weighing an external skillset or plugin against your installed skills |
convert-date |
Converting between Shamsi/Jalali and Gregorian calendars |
create-permission-hook |
Creating permission hooks for CLI tools |
create-plan-and-execute-md |
Planning and executing in one go (markdown plan) — "plan and execute markdown" |
create-plan-and-execute |
Planning and executing in one go (HTML plan) — "plan and execute", "plan and ship" |
create-plan-md |
Writing a plain markdown implementation plan — "create markdown plan" |
create-plan |
Writing a structured implementation plan to disk as interactive HTML — "create plan", "plan this" |
cut-clip |
Cutting a video/audio segment, with boundaries given as spoken phrases |
deep-research |
Researching topics — "what's the latest on X", "research X for me" |
elevenlabs-voice |
Transcribing audio (with speaker diarization) or generating speech via ElevenLabs |
exa-cli |
Exa AI neural search, grounded cited answers, clean content extraction for URLs |
execute-plan-md |
Executing a markdown plan task-by-task — "execute markdown plan" |
execute-plan |
Executing an HTML plan task-by-task with per-step verification and commits — "execute plan", "run the plan" |
fetch-raindrop-bookmarks |
Fetching and triaging Raindrop.io bookmarks |
fetch-twitter-bookmarks |
Fetching and triaging Twitter/X bookmarks |
figma-cli |
Reading Figma files — node trees, copy, rendering frames to PNG/SVG (read-only) |
find-session |
Finding a past Claude Code session by keyword and giving its resume command |
gemini-cli |
Asking Google Gemini, second opinions, managing Gemini CLI extensions/MCP servers |
generate-image |
Generating or editing an image (OpenAI GPT Image or Gemini Nano Banana) |
glab-cli |
Running GitLab operations — MRs, pipelines, issues, CI logs |
gws-cli |
Interacting with Google Workspace (Gmail, Calendar, Drive, Sheets) |
how-to-html |
Generating or substantially editing an HTML file — plans, reports, dashboards, custom-editor UIs, slides. Ships a diagram geometry verifier, a content density lint, and collapse-layer / report-nav installers |
imsg-cli |
Reading and sending iMessage/SMS on a Mac — chats, history, search |
integrate-cli |
Onboarding a new CLI end-to-end — research, install, auth, generate a <tool>-cli skill and permission hook |
learn-from |
Turning mistakes and corrections in the current conversation into edits to CLAUDE.md, memory, or skill files |
md-to-pdf |
Converting a markdown file (Obsidian-flavoured supported) to PDF |
ntn-cli |
Notion via the official ntn CLI — pages as Markdown, data sources, raw API, Workers |
post-mortem |
Writing a blameless postmortem / RCA |
python-project-setup |
Setting up new Python projects (uv + ruff + pyright + pytest) |
quick-search |
Fast single-fact web lookups: docs, syntax, versions |
refine-english |
Refining text to sound native, writing feedback, pronunciation help |
signal-cli |
Interacting with Signal — receiving, contacts, groups, sending messages |
slack-cli |
Interacting with Slack — reading, searching, sending, reactions |
snow-cli |
Running Snowflake operations — SQL queries, schema inspection, stages, Cortex |
stripe-cli |
Stripe in test mode — customers, charges, subscriptions, invoices, webhooks |
summarize-content |
Summarising a URL, YouTube video, podcast, or local media file |
tgcli |
Interacting with Telegram — reading chats, sending messages, searching |
understand-session |
Teaching and quizzing you on a session's work or a PR until you can explain it |
unleash-cli |
Unleash feature flags — state per environment, toggles, rollouts, audit log |
update-cli |
Upgrading an installed CLI and syncing its <tool>-cli skill and permission hook with new/removed commands |
update-permissions |
Adding, removing, or changing Claude Code permission rules and hooks, with consolidation and a security review |
why-you-asked-me |
Diagnosing why Claude prompted for permission on a command that should have been auto-allowed |
writing-skills |
Creating or editing a skill or a CLAUDE.md rule — triggers-only descriptions, form-to-failure, probe-first testing |
Auto-allows read-only ClickUp commands (task view, task list, task search, etc.). Write operations (task edit, comment add, status set) prompt for user confirmation.
Auto-allows read-only Google Workspace commands (+triage, +agenda, +read, message list/get). Write operations (+send, +insert, event create/update/delete) prompt for user confirmation.
Auto-allows read-only Snowflake commands (snow connection, snow object list/describe, snow stage list, snow cortex, snow logs). Write/execute operations (snow sql, create, drop, deploy, copy) prompt for user confirmation.
Auto-allows headless gemini -p prompts and read-only listings (mcp/extensions/skills list, gemma status/logs). Yolo mode (-y, --approval-mode yolo), --skip-trust, and config changes (mcp add, extensions install, gemma setup) prompt for confirmation.
Auto-allows read-only imsg commands (chats, history, search, watch, account, status). Sends, reactions, edits, deletes and chat changes prompt for confirmation.
Auto-allows read-only ntn commands (GET-style ntn api, v1/search, database queries, pages get, datasources query, whoami, doctor). Page writes, ntn api calls with a body or mutating method, file uploads, Workers ops and ntn auth token prompt for confirmation.
Auto-allows read-only signal-cli commands (list*, get*, receive, version). Every send, group/profile/device change, trust change and account operation prompts for confirmation.
Hard-blocks anything touching live mode (--live, sk_live_/rk_live_/pk_live_ keys, a live STRIPE_API_KEY). Auto-allows test-mode reads (list, retrieve, search, get, logs, listen, whoami). Test-mode writes prompt for confirmation.
Auto-allows read-only unleash wrapper commands (whoami, events, flags list, flag get/strategies, api GET). Flag create/toggle/rollout/archive and mutating api calls prompt for confirmation.
Some skills require external CLI tools. See SETUP.md for installation instructions.
| Skill | Requires |
|---|---|
babysit-pr |
gh (GitHub CLI) |
bird-cli |
bird (Twitter/X CLI) |
clickup-cli |
clickup (ClickUp CLI) |
codex-ask |
codex (OpenAI Codex CLI) |
commit-push-pr |
gh and/or glab (matches the remote) |
convert-date |
python3 with jdatetime |
cut-clip |
yt-dlp, ffmpeg, curl, ElevenLabs API key |
elevenlabs-voice |
curl, jq, python3, ElevenLabs API key |
exa-cli |
uv, Exa API key |
fetch-raindrop-bookmarks |
Raindrop.io API token |
how-to-html |
python3 3.9+ (stdlib only, for the bundled verifier, density lint and installers) |
fetch-twitter-bookmarks |
bird (Twitter/X CLI); optional yt-dlp for media |
figma-cli |
python3, Figma personal access token |
find-session |
python3 (stdlib only) |
gemini-cli |
gemini (Google Gemini CLI), GEMINI_API_KEY |
generate-image |
OPENAI_API_KEY and/or GEMINI_API_KEY |
glab-cli |
glab (GitLab CLI) |
gws-cli |
gws (Google Workspace CLI) |
helpers |
python3; claude CLI on PATH for adherence-probe.py |
imsg-cli |
imsg (iMessage CLI, macOS only) |
integrate-cli |
jq (for generated permission hooks) |
make-slides |
git |
md-to-pdf |
pandoc, Google Chrome or Chromium |
new-session |
Claude Code CLI with --bg support (or dispatch_mode: print) |
ntn-cli |
ntn (Notion CLI) |
python-project-setup |
uv |
signal-cli |
signal-cli; qrencode for device linking |
slack-cli |
agent-slack (Slack CLI) |
snow-cli |
snow (Snowflake CLI) |
stripe-cli |
stripe (Stripe CLI) |
summarize-content |
summarize CLI |
tgcli |
tgcli (Telegram CLI) |
unleash-cli |
python3, Unleash Admin API token |
See CONTRIBUTING.md for how to add new skills and the PR process.
MIT