Markdown skill files that teach an AI assistant how to do one job properly — the ones actually used to run this business. Each drops straight into Claude, ChatGPT, Hermes, or whatever you run locally. Free, editable, no signup.
📈GTM · Paid skills
A separate set of skills for ranking in Google and getting cited by ChatGPT, Perplexity and AI Overviews — keyword research, topic scoring, GEO optimization and content auditing. These ones are for sale rather than free.
Browse the store

Bundle · Search & Research
Exa
The Exa family: semantic search and content extraction over raw HTTP, the full product surface for builders, plus the company research and lead generation workflows built on top of it.
5 skills in this bundle
🧠Ideation · Creative
Parallel divergent ideation: fan five isolated agents out under different cognitive frames, score and cluster what comes back, prune the traps, then deepen the best three into buildable sketches with an actual opinion.

GTM · Data
Clay's whole programmatic surface in one reference: which of the three surfaces (REST, MCP, CLI) can actually do the thing, the non-standard auth header, what is UI-only — and the two rules that stop you spending someone's credits by accident.

Bundle · Scraping & Indexes
Firecrawl
The Firecrawl family: credentials and setup, page-level scraping, search-before-extraction, browser interactions for pages a plain scrape cannot finish, and the developer and research paper indexes.
6 skills in this bundle
🧑🔬GTM · Strategy
Six practitioner corpora — Crawford, Lieben, Voje, Hérubel, Stam, Vacca — routed by what the question is actually about. Returns their real steps and real numbers, flags where they contradict each other, and labels every claim measured, asserted or marketing.
🕵️Research · GTM
Exa finds the page that actually states the ICP, Firecrawl reads it properly, then the client logo wall gets resolved to real headcounts — so company size is observed, not guessed from a price. Every field carries a verbatim quote or stays empty.
💸AI · Cost
Before you run a model over 10,000 rows, price the job. Reads OpenRouter's live rankings and live pricing, costs your actual run in dollars, and puts the open-weight labs next to the frontier ones — the comparison a from-memory shortlist always skips.
🔭Research · Competitive
Who a company targets, not who it has already won. Reads live ads out of the LinkedIn, Meta and Google ad libraries with a real browser — they are JavaScript apps that block scrapers outright, so this is the only way in.
📊Data · API
All 17 core and values methods, all 74 batchUpdate request types, and the OAuth flow from a client ID to a working refresh token — plus the three behaviours that quietly cost people data: append writes somewhere else, reads come back ragged, and the wrong field mask erases cells.
🗄️Data · API
PostgREST, direct Postgres for bulk loads and DDL, and the four key types — including which one fails silently. Written around the three things that make inserts look successful when nothing landed: the 1000-row cap, RLS returning empty arrays instead of errors, and DDL that REST cannot do at all.
🧭Strategy · Build
Before you commit to a stack: researches how comparable projects were actually built, prices the operating cost of each option, and comes back with one recommended approach rather than a menu of them. For when there is no look-alike project to copy.
⚡Bundle · Lead Sourcing
Blitz
The Blitz pipeline in three skills: scope the GTM brief before spending a credit, turn it into a runnable script with real pagination, then preflight the integration against the live docs.
3 skills in this bundle
🧾Ops · QA
The difference between a review that happens and one deferred session after session: every machine-made claim shipped next to its source URL and a local snapshot of the text it came from, so the reviewer checks the work instead of redoing it.
💼Bundle · Signals & Leads
LinkedIn
The LinkedIn pair: turn a creator's comment sections into scored buyer leads, and verify where someone actually works before spending enrichment credits on them.
2 skills in this bundle
🧬Research · Papers
Checks arXiv before you invent an algorithm. Reads the prior art in parallel, scores and clusters it, then converges on one path with citations, a first step, and the pitfalls whoever published it already walked into.
🗂️Ops · Data
Never overwrite a file a human has already been given. The revision gets a new dated filename and the old one stays put, so the person reading yesterday's copy and the agent reading today's are never quietly looking at different data under one name.
🔒Ops · Discipline
Anti-laziness execution discipline for agents that report done at 80 percent. v2 stopped trusting promises: acceptance gates go in a file before work starts, each one carries a runnable check, and completion is proven against the ledger instead of asserted.
🎨Design · Frontend
Anti-slop frontend for landing pages, portfolios and redesigns. Reads the brief before touching a dial — page kind, vibe words, references — then pulls only the rules that fit, so the result stops looking like the template every model defaults to.