Pawel Huryn
Map an outcome to user problems, possible fixes and cheap experiments before you build
Skills you drop into Claude Code, Claude Desktop, Codex, or Cursor to teach it a job. Sort by trending, rating, or upvotes.
1037 skills
Pawel Huryn
Map an outcome to user problems, possible fixes and cheap experiments before you build
Pawel Huryn
Pick the one customer-centric number your product should move, plus the inputs that drive it
Muratcan Koylan
Design the loop around an autonomous agent: what it can edit, how it's scored, and when a human signs off
Muratcan Koylan
Fit more useful work into the same context window with budgeting, masking and cache-friendly prompts
Muratcan Koylan
Diagnose why your agent gets dumber as the session grows, and fix it before it spirals
browserbase
Port your Python browser-use scripts to TypeScript Stagehand v3 with more deterministic steps
browserbase
Scaffold a browsing agent that can only visit domains you allowlist, with no raw shell or CDP
coreyhaines31
Plan, write and repurpose posts for X, LinkedIn and friends, with the story first
coreyhaines31
Write B2B outreach that sounds like a thoughtful person, and actually lands in the inbox
coreyhaines31
Find the friction in your signup form and get more people through to their first real use
coreyhaines31
Design upgrade screens and feature gates that convert free users at the right moment
mattpocock
Give your agent a shared vocabulary for deep modules, clean seams and testable interfaces
mattpocock
Turn a folder into a learning workspace where your agent tutors you across sessions
mattpocock
A tough design interview that leaves ADRs and a glossary behind as it goes
addyosmani
Your agent asks one question at a time until it knows what you actually want built
addyosmani
Make your agent check the official docs and cite them instead of coding from stale memory
addyosmani
Turn a fuzzy idea into a sharp, buildable concept before your agent writes a line of code
Muratcan Koylan
Give your agents memory that lasts between sessions without filling it with stale facts
Muratcan Koylan
Write tool descriptions, schemas and error messages that agents can actually use well
Muratcan Koylan
Decide if you really need more than one agent, and if so, how to split the work cleanly
Muratcan Koylan
Keep long agent sessions on track by summarizing context without losing the decisions that matter
Pawel Huryn
Turn tickets, git logs or a messy changelog into release notes people will actually read
Pawel Huryn
Imagine your launch flopped, then work backwards to the risks that actually matter
Pawel Huryn
Get a lean, one-to-two page PRD your team can actually decide on, drafted by your agent