Plans an install ad that only claims what the product actually did.
Early AccessEvery agent is free to connect — no card, no checkout.
We use essential cookies to keep you signed in. With your consent we also use non-essential cookies for analytics. Read our privacy policy
Browse › Marketing & Growth › Marketing Hook Writer
Writes hooks in the customer's own words, checked against what the page can promise.
A preview derived from this agent’s definition — not a live run. Nothing runs and no key is sent until you connect. FindAgent runs it on the hosted connection.
Pull the exact customer phrases that will be the raw material for every hook, kept word for word with their sources, and judge whether they survive the campaign's creative language.
Write the hooks for one angle as three separable parts — spoken, on-screen, and what is happening — covering distinct archetypes rather than rephrasing one idea.
Write the body skeletons, proof modules and calls to action for an angle as separable units, each proof module naming the asset it needs and whether it exists.
Check written hooks against the claims boundary, the banned list and the angle's do-not-say, flagging each line that cannot ship and why.
Lay out how the modules combine into testable variants, so each cell differs from its neighbour by exactly one variable and the result can be attributed.
Assemble the creative bank for the script stage — hooks, bodies, proofs, CTAs, the matrix and the rejected lines — refusing to produce one when no modules were written.
This MCP server runs on your own machine — FindAgent only points you at it. Run the command below, then add it to your MCP client. Nothing is hosted or proxied on our side.
npx -y @findagent/mcp runPaste this into your client config (e.g. claude_desktop_config.json or a Cursor mcp.json), then restart the client.
{
"mcpServers": {
"marketing-hook-writer": {
"command": "npx",
"args": [
"-y",
"@findagent/mcp",
"run"
]
}
}
}Plans an install ad that only claims what the product actually did.
Turns YouTube Analytics metrics into a deterministic performance narrative + recommendations
A 0–100 site health score from GA4, per device, with the failing metrics and what to do about each.
Scores only what was measured, and refuses to read a test that cannot be read.
Ranks competitor ads by how long they survived, and says what it could not reach.
Scores ad angles with arithmetic you can audit, not a ranking you have to trust.
Flags the days your GA4 metrics deviate from their baseline — with the numbers computed, not guessed.
Maps your module graph, scores blast radius, plans migrations and composes a validated unified diff.
Turns a vague mobile feature request into a spec with a real scope boundary.