Find the next skill your agents load.

Use when #customer-escalations contains a ticket link plus “needs repro.”

View skill

Fathom poll → summary + two action lists

by @p0lybender

Use when meetings are recorded in Fathom and the human needs you-committed vs they-committed lists.

View skill

Gmail triage — draft only, never send

by @p0lybender

Use when the 7:30am Gmail charter runs: easy-archive, glance mail, action-mail drafts.

View skill

2–6 Bot group with one owner per stage

Use when Research → Write → Review must be a group, not a silent DM pile.

View skill

Pull two months of open items → Sheet Kanban → daily nudge

by @rajesh_thallam

Use when tasks leak across notes, chat, email, docs and standups need one board.

View skill

standup-notes SKILL.md: git log last 24h → copy-paste standup draft

by DeployHQ

Use when the human needs this morning's standup and the repo is the source.

View skill

Decide

anthropics/skills webapp-testing — Playwright + with_server.py black box

by spencerpauly/awesome-cursor-skills

Use when verifying frontend, screenshots, or browser logs against a local server.

View skill

Claude Code /aun SKILL.md — one AUN URL, AskUserQuestion, no '対応済'

by @AUN_PHONOGRAM

Use when the human says fix this AUN /aun URL and the repo files must match the comments.

View skill

Babysitting-PR SKILL.md: keep an open PR merge-ready

Use when a PR is open and the agent should keep it green without merging.

View skill

Claude Code hooks: Prettier, protect-files, Stop prompt, git if-filter

by Ajit Singh (singhajit.com)

Use when Prettier, protect-files, Stop prompt, or git if-filter must always run.

View skill

First SKILL.md: commit-helper + slash-only prod-deploy

Use when commits can auto-draft but production deploy must only run on /prod-deploy.

View skill

Project hooks.json: afterFileEdit formatter + cloud-safe command hooks

by Cursor (cursor.com/docs/hooks)

Use when every edit should run a repo-root formatter hook.

View skill

disable-model-invocation for deploy/PR/billing skills

by Cursor docs

Use when a skill would send, pay, deploy, force-push, or merge if the model auto-loaded it.

View skill

Draft-first Auto-review + takeover; Bots are not a security boundary

by Nous Research Hermes Agent docs

Use when setting stop lines for send, pay, deploy, delete, and Auto-review rules.

View skill

Grok writes Hermes Kanban cards sized to context

by @tmophoto

Use when Grok plans, Hermes executes, Grok reviews, and cards must fit 6–8 concurrent tasks.

View skill

HAEGONG Grok Bot profiles: Spec Writer → human approval → PR Producer; PR Verifier in a separate chat

by HAEGONG

Use when a coding change needs a spec gate before any PR and a verifier that did not write the PR.

View skill

Hermes Kanban — decompose, don’t execute

Use when a board should break jobs into cards for other agents, not run the code.

View skill

Solo-founder Grok Bots: interview once, state file, if-nothing-send-nothing, Outer Loop never commits

by Paul Klay (2026-08-29 “How to Build the Best Team for Solo Founders in Grok Bot”)

Use when a quiet night is the correct output and a chatty bot would wake the human.

View skill

Fable Forensic Skill — verify the artifact, not the performance

by @kuroi_kuropug

Use when a demo, eval, or article might be theater and the file is the evidence.

View skill

loop-runner SKILL.md: one iteration, files are the only memory, cron owns cadence

by frankxai

Use when a recurring job should not remember in chat.

View skill

Claude Fable 5.1 deep-article SKILL.md (effort:high, fork research)

by @MakeAI_CEO

Use when a long research article is the output, not a pulse blurb.

View skill

obra/superpowers brainstorming — design-before-code gate

by @obra

Use when a coding task is about to start without a written design.

View skill

obra/superpowers test-driven-development skill (Iron Law)

by @obra

Use when implementation is about to start without a failing test.

View skill

memory-core dreaming: 03:00 cron, light→REM→deep; only deep writes MEMORY.md

by Akshay Pachaar (quoted from 2026-08-26 “Build an Engineering Team With Grok Bot”; synthesis by Shirley / AI Builder Club)

Use when daily notes should compact into MEMORY.md overnight.

View skill

Four-Bot engineering team: on-disk board, one writer, Blocked not guessed

Use when parallel engineering bots would otherwise collide on the same files.

View skill

Loops: Grok Bot as outer loop that fires Cursor cloud agents

by Matt Palmer (Cursor / SpaceXAI; “Chat is all you need” 2026-08-18; Loops template 2026-08-28)

Use when the manager Bot should not do the compute.

View skill

Python TDD with uv SKILL.md (red-green-refactor, one failing test)

by spencerpauly/awesome-cursor-skills (cites mattpocock/skills TDD)

Use when writing or fixing Python and a uv-based TDD skill is installed.

View skill

/execute-plan: one-commit stages, wait for go, never push

by Roman Zipp (2026-08-17)

Use when a plan is approved but each stage still needs a human go.

View skill

Grok Bot coordinator + HANDOFF block (never send/post/buy/delete)

by @zodchiii

Use when work moves between bots as a HANDOFF block on disk.

View skill

Outbound

LinkedIn engager → waterfall enrich → Instantly

by Flavio Copes (updated 2026-08-30)

Use when a buyer-matching LinkedIn post has engagers and the list is not yet in Instantly.

View skill

Five-bot outbound team that cannot send

by @coldemailchris

Use when one agent covering find/watch/write/inbox is averaging four jobs and mistakes are too big.

View skill

Blog Pulse: write the role, run once, save skill, then schedule with no-guess gaps

Use when the site needs an 08:00 pulse and the bot must not publish.

View skill

Build a send-voice guide from outbox + Slack, then sample the thread

by SpaceXAI Grok Bot Guides (Aug 15, 2026)

Use before any outbound draft — agents write like you only after they read you.

View skill

Force x_search + web_search in parallel, never default-web

by AIToolsRecap (Jun 16, 2026)

Use when live X firehose and indexed web both matter (competitive, mentions, pricing).

View skill

Grow

AI search is SEO (best X for Y)

Use when the page must show up in AI answers by first ranking pages 1–3 of classic results.

View skill

Claude Code end-to-end Meta lead-gen (browser BM + API, never ACTIVE)

by @boringmarketer

Use when the Meta campaign must be created end-to-end and the agent must not spend money.

View skill

Meta Ads MCP skill — create PAUSED, never delete, never +20% budget

by Digitizers

Use when the Facebook Ads MCP is connected and the user mentions Meta campaigns, budgets, or pixel.

View skill

Facebook Advantage+ outcome ads

Use when Meta creative should sell the customer's desired outcome, not tinker targeting.

View skill

OpenClaw meta-ads SKILL.md — daily ops + 8-stage /meta-campaign, ads paused

by fdarkaou

Use when a landing-page URL should become a paused Meta campaign, or a morning brief should run dry-run first.

View skill

Google Ads BOFU keyword-family agent

Use when the account can fire a deep conversion and keywords need family-level ad sets.

View skill

Replace stale links with a better page

Use when lots of sites already link a weaker page you can beat.

View skill

LinkedIn remix → thought-leadership ads

Use when organic remix should become a LinkedIn ad with a signup CTA and engager capture.

View skill

Marketing agent access stack

Use at install, before Google/Meta/Instantly skills.

View skill

SEO/AEO workflow — GSC → keywords → CMS → draft for approval

by @startupideaspod

Use when Search Console and Ahrefs/SEMrush should become a post, meta, and internal links — not a publish.

View skill

BOFU SEO + monthly refresh

by Anthropic / Agent Skills spec

Use for X vs Y, X alternative, X review, how to X pages you can own.

View skill

Keep

Write every agent spec like a job description

by @startupideaspod

Use when a new agent or skill needs a hire-style spec before any routine.

View skill

Official Agent Skills SKILL.md spec (agentskills.io)

Use when packaging a skill directory with required SKILL.md plus optional scripts/references/assets.

View skill

AGENTS.md nested playbook + Cursor Cloud specific instructions

by Cursor (docs/rules + docs/cloud-agent/setup + docs/cloud-agent/best-practices)

Use when placing AGENTS.md at repo root and in nested packages so closer files win.

View skill

anthropics/skills mcp-builder — four-phase MCP server skill

Use when an LLM must call an external API through well-named tools, not a one-off script.

View skill

anthropics/skills pdf — production document skill with FORMS.md split

Use when any PDF extract, merge, form-fill, or OCR is the job — trigger on .pdf, not on the word skill.

View skill

anthropics/skills skill-creator — write, eval, iterate SKILL.md

Use when a new skill needs write → eval → iterate, or an existing description needs better triggers.

View skill

Claude Code SKILL.md: invoke control, !shell inject, fork, allowed-tools

by Justin Trugman, BetterFutureLabs (Jun 15, 2026)

Use when a Claude Code skill needs disable-model-invocation, fork, or allowed-tools beyond the spec six.

View skill

GitHub plugin marketplace so skills leave one laptop

Use when a team needs /plugin marketplace add org/repo so skills version together.

View skill

Claude Code cron: /loop, CronCreate, Desktop tasks, cloud Routines

Use when a skill is trusted and the next question is which scheduler owns cadence.

View skill

Claude Code memory: CLAUDE.md, .claude/rules, auto MEMORY.md

by Cursor (docs/cloud-agent/setup + docs/cloud-agent/builds)

Use when instructions and learnings must load at session start as files, not chat.

View skill

Cloud Agent environment.json: install vs start vs terminals + Dockerfile

Use when a Cloud Agent Build must be shared and long-running processes must survive the snapshot.

View skill

Cursor Automations: /automate, cron, MEMORIES.md, PR/Slack tools

by Cursor (docs/cloud-agent/automations + help/ai-features/automations)

Use after the skill is trusted and a trigger (cron, PR, Slack) should fire it.

View skill

Project rules as .cursor/rules/*.mdc (always / globs / intelligent / manual)

by Cursor (cursor.com/docs)

Use when the team needs versioned rules that are not a SKILL.md procedure.

View skill

Official Cursor SKILL.md format, discovery paths, and built-in /create-skill

by Cursor (cursor.com/docs)

Use when a Cursor skill must live in .cursor/skills/ (or .agents/skills/) with name matching the folder.

View skill

Package a Cursor SKILL.md (discover → activate → scripts)

by Cursor docs

Use when a workflow needs a folder with SKILL.md, then references/ and scripts/ on demand.

View skill

Named bots manage. Heavy Lift Cloud Agents do the compute.

by @thedogfather

Use when a CoS or specialist is about to code, browse long, or process data on the Grok Bot computer.

View skill

Named Bots, scoped memory, smallest useful roster

by SpaceXAI Grok Bot Guides (Aug 15, 2026)

Use when one omniscient agent is mixing mail, evals, and outbound in the same memory.

View skill

Install skills via plugin folders + marketplace.json

by @grok

Use when SKILL.md packages should load the same way plugins do.

View skill

One shared computer, /workspace handoffs, screens ≠ security

Use when two Bots must share files, cookies, or a login.

View skill

Start with SKILL.md, watch the browser, refine

by @cicerougc

Use when making a browser agent and the skill does not yet hold.

View skill

Skill runtime — load, execute, unload + approval field

by @p0lybender

Use when writing or reviewing a skill's six fields — especially the approval field people skip.

View skill

A Grok Bot skill is a SKILL.md

by @joingroklab

Use when someone is about to paste a giant prompt instead of a reusable recipe.

View skill

Task → 6-field skill → routine (never reverse)

Use when the first good output is about to be scheduled.

View skill

Grok Bot Super: URL playbooks → official save-as-skill

by AgentMindCloud (MIT)

Use when a public playbook URL is the source and save-as-skill is the output.

View skill

Teach a task — 10-minute browser demo → draft skill

by Khalid Abdelaty (DataCamp); architecture from docs.x.ai/grok-bot/computer-and-apps and docs.x.ai/grok-bot/bots

Use when Teach a task is available and the workflow is visible on the computer.

View skill

Grok Bot Templates — package skills, memory, plugins as a link

by @ATPinsights

Use when a taught workflow should be a blueprint others can install.

View skill

Grok Bot: put changing facts in /workspace, not Bot memory

Use when a Bot is about to 'remember' a fact that will go stale in the profile.

View skill

Grok Build CLI: grok -p headless, CI, sessions

by xAI / SpaceXAI

Use when a trusted skill should run non-interactively in CI.

View skill

Grok Build: SKILL.md layout, /skillify, AGENTS.md, /loop

by xAI / SpaceXAI; /skillify from x.ai/cli; /loop from scriptbyai.com cheatsheet (audited Grok Build v1.0.3, 2026-08-15)

Use after a one-off Build task worked and the method should be reusable.

View skill

Hermes cron: skill-backed jobs, chaining, no-agent watchdogs

by Nous Research (Hermes Agent docs)

Use when a trusted Hermes skill should run on a clock without improvising.

View skill

Hermes SKILL.md — delegate coding to grok -p (no --always-approve on audits)

by Matt Maximo (MattMaximo) / Hermes Agent

Use when the Hermes agent owns orchestration and Grok Build owns the code.

View skill

Hermes SKILL.md hardline + blueprint cron (opt-in /suggestions)

by NousResearch / Hermes Agent

Use when a Hermes skill must meet the hardline layout and cron is opt-in via /suggestions.

View skill

Hermes bounded memory: MEMORY.md + USER.md

by Nous Research (Hermes Agent docs)

Use when Hermes is about to accumulate unbounded chat memory.

View skill

Hermes Agent: SKILL.md layout, /learn, skill_manage, hub

by Nous Research (Hermes Agent docs)

Use when a Hermes skill must match hub layout so /learn and skillmanage can find it.

View skill

Chief of staff runs 10–20 named Grok Bots

by @coldemailchris

Use at install: smallest useful roster plus a CoS, not 20 bots on day one.

View skill

Install skills from ClawHub, git:owner/repo, or a local SKILL.md dir

Use when a skill is not bundled and must be installed from hub or git.

View skill

Isolated automations: fresh cron: session, NO_REPLY if idle, announce to a channel

Use when a trusted skill should fire isolated from the main chat.

View skill

OpenClaw GitHub skill — slash install, PAT in vault, then gh

by @aiclawbots

Use when the agent needs gh and the PAT must not sit in a dotenv.

View skill

OpenClaw memory is files: USER.md, MEMORY.md, memory/YYYY-MM-DD.md, DREAMS.md

Use when OpenClaw is about to remember in chat instead of the file layout.

View skill

Bundled skill-creator: contract → invocation → SKILL.md + scripts/ → workshop → validate

by openclaw/openclaw

Use when a new OpenClaw skill must be created to the bundled contract.

View skill

SKILL.md + metadata.openclaw gates (bins/env/config) or it never loads

Use when an OpenClaw skill is installed but silent because gates failed.

View skill

Skill Workshop: PROPOSAL.md first, apply is the only live write

by S Sid Bharath (Refound AI)

Use when capturing or editing an OpenClaw skill through the workshop.

View skill

Grok Bot Chief of Staff that reads COMPANY.md / APPROVALS.md first

Use at every CoS turn before routing work.

View skill

Saving workspace context: file-based agent memory across chats

by spencerpauly/awesome-cursor-skills

Use when a new chat would otherwise forget the last session's state.

View skill

When the agent spins — ask why, then save the skill

by @mikeassad77

Use at the moment the agent is slow or looping, not in the abstract.

View skill

vercel-labs/agent-skills react-best-practices (70 rules, progressive)

by Vercel (@vercel / vercel-labs)

Use when writing or reviewing React/Next UI and the skill is installed.

View skill

Look-lock before any mesh

Use when a 3D, game, or scene build is about to start without locked reference stills.

View skill

Screenshot-match until look passes

Use when a build must match locked look plates via screenshot → delta → fix loops.

View skill

Blender MCP scene out

Use when blocking, materials, camera, or render should be driven through Blender MCP (not pasted Python).

View skill

Computer-use 3D artist (GUI drive)

Use when Blender/Unreal/Unity must be driven via desktop GUI because MCP is unavailable.

View skill

Photo → editable 3D rebuild

Use when a user photo or reference must become an editable Blender scene matched back to the photo.

View skill

Concept → Blender → engine pipe

Use when a concept still must become a graybox in Blender then a playable slice in Unreal or Unity.

View skill

Floorplan → walkthrough

Use when a floorplan image or PDF must become a scaled 3D walkthrough or orbit scene.

View skill

Experience-first game brief

Use before art or tech when a game idea needs player fantasy, 30s loop, win/lose, controls, and one unique verb.

View skill

Agent playtest harness

Use when a build needs scripted agent play sessions with screenshots, logs, and a bug list with repro.

View skill

Parallel asset swarm