Tools provided
Tools the agent exposes — your AI client calls them automatically when it needs them.
- get_listing_style_guide — Returns the FindAgent listing-asset rules: the workflow, the logo family, screenshot rules, the template catalogue and the title/tagline limits. Read it before planning assets.
- list_themes — Lists the looks (themes), screenshot concepts (art direction) and logo styles. A theme sets the logo colours and the screenshot ground and is picked from the agent name, then its category, unless named; the logo style defaults to the theme's own. Use it to give each agent a look that matches what it does.
- list_logo_symbols — Lists the built-in logo symbols with the kind of agent each one suits.
- compare_logo_options — Renders 2 to 4 logo candidates side by side at 160, 64 and 32 px on light and dark grounds, as one PNG, so the owner can pick one. Candidates may differ in symbol, theme or style, so you can offer different looks, not just different symbols.
- render_logo — Renders the chosen logo as PNG (transparent corners) at the requested sizes, plus the SVG source, in the theme and style that fit the agent (picked from agent_name and category unless given).
- render_screenshot — Renders one 1600x1000 marketing screenshot: themed ground, a bold headline (*asterisks* mark the accent part, \n breaks a line), optional subhead and corner badge, logo and name, and one visual. Scenes show the agent in its medium (phones, chat, code, document, dashboard, timeline, queue); diagrams explain it (transform, question, checklist, ranked, split). Data shapes are in the data field. Warns when text was cut, or would not read on a gallery card.
- check_listing_copy — Checks a planned title, tagline and screenshot copy against the listing limits and the source page: lengths, headline and subhead words, a subhead that repeats the tagline, code-like tool names, claims and numbers the page does not back, and one idea repeated across shots. Reports only; it never rewrites.
- open_form — Use this the moment the user wants listing assets made. It returns the questions to ask (agent name, logo symbol, the screenshots, and optionally title, tagline and page copy) so a form can be shown. Nothing is rendered yet.
- run_form — Internal: invoked when the user submits the form opened by open_form. Not for direct use.
- run_full — Make every listing asset in one call: check the title, tagline and headlines against the limits and the page copy, render the logo (1024 and 512 px PNG plus SVG) and render each screenshot (1600x1000 PNG). Use when the logo and the screenshot plan are decided. Missing input, it asks for it instead of failing.
- list_capabilities — List what this agent makes, the templates and symbols, the copy-check rules and its limits (no network, no credentials). Use for discovery before planning assets.
- plan_inputs — Plan the inputs for a tool before running it: returns the questions, the schema and a ready-to-edit example. Use when the user is unsure what to provide. tool is one of run_full, render_logo, compare_logo_options, render_screenshot, check_listing_copy.
- discover_intent — Use when the user states a goal in their own words ('make my listing look like the others'). It restates the goal, asks the few questions that pin the input down and proposes a ready-to-run input to confirm before rendering.
