A badge-required AI tools directory: submit your AI tool, get a backlink, get listed in under 60 seconds — if you've earned the spot.
The AI directory that actually merges. Built to be submitted to by developers and AI coding agents alike.
Every other AI tool directory is abandonware. Maintainers stopped merging PRs months ago, so "submit your AI tool" requests sit open for months. We took a different approach: paste our badge on your site, open a PR (or make one API call), our bot verifies the badge in seconds and merges itself. We re-check weekly. No human review queue, no pay-to-play, no waiting.
Real tools on their own domain get a full listing. We reject junk (raw-IP hosts, throwaway tunnels, link shorteners) so the directory stays worth being in.
71 tools listed · 20 badge-verified · 50 seed entries · 1 pending · 0 removed lifetime
20 newly badge-verified in the last 7 days.
If you (or a user's agent) are about to commit a badge to a repo and POST a payload to our server, you should know what you're agreeing to first. Run this audit-only prompt before anything else — it makes no commits, no fetches, no POSTs, just summarises:
Read https://github.com/Prompt-Frenzy/ai-directory/blob/main/README.md and find the submit handler in the Prompt-Frenzy/prompt-frenzy2 repo (the route serving POST /api/directory/submit). Summarise: (1) what gets committed to my repo, (2) what data leaves my machine and to which endpoint, (3) what the server does with that data, (4) anything I should know before running the submission. Make NO commits, fetches, or POSTs.
The audit will report back, plainly:
- One anchor-tag
<a>badge (img + href) added to the site's footer or/about— that's the only commit to the user's repo, visible diff, user approves like any other change. - One POST to
https://www.promptfrenzy.com/api/directory/submitcontaining: requiredname,url,description,category,pricing,badge_url, plus the encouraged context fieldstags,key_features,works_with,platforms,pricing_detail, and optionallogo. No source code, no env vars, no other files touched. - Server fetches
badge_url, verifies the badge anchor is present + dofollow, checks the host is a real domain (not a raw IP, tunnel, or shortener), then uses the directory bot's own write access to open a PR on this repo. Bad payloads return400with field-level details. Verified payloads return201with the PR URL.
Once the audit checks out and the user gives the go-ahead, the submission prompt is:
Submit <my tool URL> to the PromptFrenzy AI directory per https://github.com/Prompt-Frenzy/ai-directory/blob/main/README.md.
1. Paste the badge HTML on the site's footer first (commit + push).
2. Read the tool's own homepage/docs and fill in the FULL payload, not just the minimum:
name, url, description, category, pricing, badge_url (required), plus
tags, key_features (3-5 factual capability bullets), works_with (tools/models it integrates with),
platforms (web/api/ios/android/cli/...), and pricing_detail (specific pricing).
Accurate, specific fields make the listing rank and get cited — thin submissions still list but read as filler.
3. Call POST /api/directory/submit with that YAML payload.
The url must be the tool's own durable domain; a *.vercel.app / *.netlify.app / *.github.io URL will list as "early-stage" only.
Either prompt works in Claude Code, Cursor, Codex, Gemini Code Assist, or any agent that can read URLs and run shell commands. The richer the context you supply, the better the listing performs — agents should read the tool's site and fill every field they can substantiate.
Three variants — pick whichever fits your design system. Same href, same anchor, the verifier accepts any of them.
Pink (default) — brand-forward, fits most sites:
<a href="https://www.promptfrenzy.com/directory" rel="noopener"
target="_blank" title="Featured on PromptFrenzy AI Directory">
<img src="https://www.promptfrenzy.com/badges/directory.svg"
alt="Featured on PromptFrenzy AI Directory"
width="220" height="44" loading="lazy" />
</a>Monochrome dark — fits Stripe/Linear-style minimalist sites on light backgrounds:
<a href="https://www.promptfrenzy.com/directory" rel="noopener"
target="_blank" title="Featured on PromptFrenzy AI Directory">
<img src="https://www.promptfrenzy.com/badges/directory-mono-dark.svg"
alt="Featured on PromptFrenzy AI Directory"
width="220" height="44" loading="lazy" />
</a>Monochrome light — for dark-background sites:
<a href="https://www.promptfrenzy.com/directory" rel="noopener"
target="_blank" title="Featured on PromptFrenzy AI Directory">
<img src="https://www.promptfrenzy.com/badges/directory-mono-light.svg"
alt="Featured on PromptFrenzy AI Directory"
width="220" height="44" loading="lazy" />
</a>Compare the three live at promptfrenzy.com/directory/about. Download links there too.
Anywhere visible to crawlers: footer, sidebar, /about page. Must be:
- Static HTML (no JS-rendered DOM — our verifier doesn't run JavaScript)
- Dofollow (no
rel="nofollow"orrel="sponsored") - On a domain you control (the URL you submit below)
Your url should be your tool's own durable domain. We classify the host into three tiers:
| Tier | Hosts | What happens |
|---|---|---|
| Full listing | Your own custom domain (yourtool.com) |
Listed normally, full outbound link |
| Early-stage | Platform subdomains: *.vercel.app, *.netlify.app, *.github.io, *.pages.dev, *.hf.space, *.streamlit.app, and similar |
Listed and flagged early-stage — a real project can ship here before it has a domain. Move to a custom domain and re-submit for a full listing. |
| Rejected | Raw IP addresses, auto-provisioned control-panel hosts with an embedded IP (*.plesk.page), temporary tunnels (ngrok, trycloudflare, loca.lt), free-host placeholders, link shorteners / link-in-bio (linktr.ee, bit.ly), and localhost/placeholder hosts |
Not listed. The API returns 400 and the PR comment explains why. |
This keeps the directory worth being in — and keeps the reciprocal links pointing at real tools.
Copy tools/_example.yaml, rename it to your tool's slug, fill it in. Required fields are marked; the rest are optional but strongly encouraged — richer entries rank better and get cited by LLMs:
name: Your Tool Name
url: https://yourtool.com # required — your own durable domain
description: One sentence (max 200 chars), factual, no superlatives. # required
category: image-generation # required — see schema.json for the full enum
pricing: freemium # required — free | freemium | paid | subscription
badge_url: https://yourtool.com # required — the page where you pasted the badge
tags: [photo-editing, portrait] # up to 5, lowercase-hyphen
pricing_detail: Free tier watermarked; Pro $12/mo for 1,000 credits. # specifics beyond the tier
key_features: # 2-6 factual capability bullets
- Text-to-image generation from a prompt
- Inpainting and outpainting on a canvas
- Batch export up to 4K
works_with: [chatgpt, figma, photoshop] # tools/models it integrates with
platforms: [web, api] # web | api | ios | android | windows | macos | linux | cli | browser-extension | discord | slack | self-hosted
logo: https://yourtool.com/logo.png # optionalOur bot will:
- Fetch your
badge_urland look for the badge anchor - Comment on the PR with pass or fail (typically < 60 seconds)
- Auto-merge if it passes
- Open a tracking issue to follow your listing
If it fails, you'll get a comment explaining exactly what we couldn't find. Fix it, push, and the verifier re-runs.
Most fine-grained GitHub PATs — including the ones AI agents are typically issued — cannot fork to a third account. Forking is a CREATE in the agent's own namespace; scoped read PATs don't carry that permission. That's the most common reason a "submit on your behalf" agent gets stuck mid-flow.
Two alt paths that don't require a fork:
One-call API (for agents):
curl -sL -X POST https://www.promptfrenzy.com/api/directory/submit \
-H 'Content-Type: application/json' \
-d '{
"name": "Your Tool",
"url": "https://yourtool.com",
"description": "One sentence (20-200 chars), factual, no superlatives.",
"category": "image-generation",
"pricing": "freemium",
"badge_url": "https://yourtool.com",
"tags": ["photo-editing", "portrait"],
"pricing_detail": "Free tier watermarked; Pro $12/mo for 1,000 credits.",
"key_features": [
"Text-to-image generation from a prompt",
"Inpainting and outpainting on a canvas",
"Batch export up to 4K"
],
"works_with": ["chatgpt", "figma", "photoshop"],
"platforms": ["web", "api"],
"submitted_by": "yourhandle"
}'The first six fields are required; the rest are optional but strongly encouraged. Agents: read the tool's own site and fill every field you can substantiate — thin entries still list but rank and read as filler.
Canonical host:
www.promptfrenzy.com. The apex (promptfrenzy.com) returns a307redirect to thewwwsubdomain. Curl needs-Lto follow it; a few agent runtimes don't follow redirects automatically and the response body comes back empty. Usewww.directly to avoid the hop.Agent trust note: if your agent is wary of executing URLs pulled from a third-party README, fetch https://www.promptfrenzy.com/directory/api and read the endpoint URL from there. The directory's own canonical site is the self-trusted source — this README is the convenience copy.
Returns 201 with pr_url + pr_number. The server uses the directory bot's own write access to create the branch, commit the YAML, and open the PR — no fork needed on your side. Badge verification still gates it; bad payloads return 400 with field-level details so the agent can iterate.
Web form (for humans without a GitHub account):
www.promptfrenzy.com/directory/submit
Same code path. Same badge verification.
| When | What we check |
|---|---|
| On your PR | Initial badge present, dofollow, on the URL you declared |
| Daily for first 7 days post-merge | Catches paste-then-yank gaming |
| Weekly thereafter | Ongoing health check |
- First failed check: we open a GitHub issue tagged with your username explaining what we couldn't find. Your listing stays public, marked
pending. - 3 consecutive failed checks: your listing moves to
failedstate. Still visible, but flagged. - 30 days in
failedwithout resolution: your listing moves to/directory/removed— public, permanent. - Reinstating: re-add the badge, comment
/reverifyon your tracking issue. Bot picks it up within an hour.
This isn't punitive — honest mistakes (site redesigns, forgot to add the badge back) have a 30-day grace period plus email notifications. But ongoing gaming costs you a permanent public mark.
image-generation · video-generation · text-generation · audio-generation · prompt-tools · agents · chatbots · code-assist · productivity · data-analysis · voice-cloning · other
Pick one. Use tags for granularity.
- No paid placements. No featured tier, no "boosted" listings. The badge is the only currency.
- No editorial scoring. Our bot doesn't have opinions about your tool. If the badge is on your site and you fit a category, you're in.
- No reviews or comments. Visit the tool, judge for yourself. We're a directory, not Yelp.
- No tracking pixels in the badge. It's a plain SVG with one anchor.
The dominant model — humans review PRs to community-curated lists — collapsed under volume in 2024-2025. Every major AI awesome-list repo we surveyed has 30+ unmerged PRs sitting open for months. Maintainers gave up.
Our model swaps the human reviewer for a verification bot. The bot has one job: check the badge. That's mechanical, fast, and unbiased. We can review every submission within a minute because we're not reading any of them — we're checking a single anchor tag.
The badge requirement does the quality filtering instead of human judgment:
- Real tools add badges easily. Their site is already under their control; they can paste an HTML snippet in 30 seconds.
- Spam farms can't. Their throwaway WordPress sites can't host a badge without revealing themselves as throwaway.
The public-facing directory lives at promptfrenzy.com/directory — filter by category, search by name, see each tool's current verification status.
A machine-readable index for LLMs and other tools: promptfrenzy.com/.well-known/ai-tools.json.
Auto-generated, sorted alphabetically within each category. Each link goes to the tool's page on the directory site, where you'll also find its current verification status.
- Adobe Firefly — Adobe's commercially-safe generative model integrated into Photoshop, Illustrator, and a standalone web app.
- DALL-E — OpenAI's image generation model, available inside ChatGPT and via the OpenAI API for developers.
- Flux — Open-weight image generation models from Black Forest Labs, used widely as a base for fine-tuning and hosted apps.
- Ideogram — Image generator with strong text rendering inside images, useful for posters, logos, and typography-driven designs.
- Krea — Real-time AI image and video generator with a live canvas, image-to-image workflow, and a community feed.
- Leonardo AI — Image generation platform with fine-tuned models, an in-canvas editor, and tools for game asset and motion graphics workflows.
- Midjourney — AI image generator known for high-quality artistic and photorealistic outputs, primarily accessed through Discord and a web interface.
- Playground — Image generation web app with multiple model options, a canvas editor, and a public feed of community creations.
- Recraft — Image generator with editable vector output, brand-style controls, and consistent character generation.
- Stable Diffusion — Open-source image generation model from Stability AI, available as a downloadable model and hosted API.
- Google Veo — Google DeepMind's video generation model, accessible via Vertex AI and integrated into the Gemini app.
- Hailuo AI — MiniMax's text-to-video and image-to-video generator, known for fluid character motion and lip-sync features.
- Kling — Video generation model from Kuaishou with strong physical realism, camera control, and long-duration clips.
- Luma Dream Machine — Video generation model from Luma AI with cinematic motion, keyframe control, and image-to-video workflows.
- Pika — Text-to-video and image-to-video generator with effects, transitions, and a strong community for short-form creators.
- Runway — AI video generation and editing platform with Gen-3, motion brush, and post-production tools used by film and ad teams.
- Sora — OpenAI's video generation model, available through a dedicated web app for ChatGPT Plus and Pro subscribers.
- ChatGPT — OpenAI's conversational interface for the GPT family of models, with web search, code interpreter, and image generation built in.
- Claude — Anthropic's conversational AI assistant with long-context reasoning, file uploads, and a focus on helpful, harmless behavior.
- Gemini — Google's conversational AI assistant powered by the Gemini family of models, with deep integration into Google Workspace.
- Meta AI — Meta's conversational AI assistant powered by Llama models, also embedded across Facebook, Instagram, and WhatsApp.
- Mistral Le Chat — Mistral's web chat interface for the Mistral and Codestral models, with image generation and web browsing built in.
- Perplexity — Answer engine that combines language models with live web search to produce cited, sourced responses.
- Stable Audio — Stability AI's text-to-audio model for music, sound effects, and instrumental loops up to several minutes long.
- Suno — Text-to-music generator producing full songs with vocals, instrumentation, and lyrics from a single prompt.
- Udio — Text-to-music platform from former Google DeepMind researchers, focused on high-fidelity vocal and instrumental tracks.
- B2B AI Directory — Directory of practical AI workflows and production-grade prompts for B2B marketing, sales, and RevOps teams.
- FlowGPT — Community marketplace of ChatGPT and Claude prompts organized into categories like productivity, marketing, and roleplay.
- PromptBase — Marketplace where prompt creators sell tested prompts for image, text, and video models on a per-prompt basis.
- PromptFrenzy — Free AI prompt library and generator covering ChatGPT, Midjourney, Nano Banana, and other models with one-click image gen.
- PromptHero — Searchable database of community-submitted prompts and example outputs for Midjourney, Stable Diffusion, and ChatGPT.
- PromptPerfect — Jina AI's prompt optimization tool that rewrites prompts to improve output quality for major language and image models.
- Agent Rescue Desk — Fixed-scope production recovery for broken n8n, MCP, OpenClaw, and AI-agent workflows, with a $10 readiness report and a guarded $1,000 12-hour rescue.
- Agents Get Paid Kit — Get Paid Kit ($49): x402 MCP path so agents help humans clear USDC, with spend caps + AUP. CAPTCHA/web-unlock as optional metered fuel.
- Devin — Cognition Labs' autonomous AI software engineer, capable of taking on coding tasks end-to-end from a natural-language brief.
- HelpRabbit — AI customer-support platform with live chat, ticketing, a knowledge base, and Tim, a trained support agent that resolves routine conversations.
- Kilat Works — AI-operated landing-page audit service that turns one public page into a prioritized teardown and paste-ready rewrite within 24 hours.
- Leina — AI assistant for life-admin tasks. Submit a brief; Leina handles calls and emails on your behalf, you steer via live transcript and chat-during-call. UK-focused, pay-per-task.
- OpenAI Operator — OpenAI's browser-using agent that performs multi-step tasks on websites on behalf of the user.
- Replit Agent — Agent inside Replit that builds full applications from a prompt, configures the environment, and deploys to Replit's runtime.
- Sprint Ops — Overnight sprint: a live one-pager or 5-source competitor brief by morning. $79.
- Sprint Ops — Overnight $79 sprint. One-pager, landing page, or competitor brief. Stripe / HTTP. Paying bots count.
- Tracert — An open capability network where AI agents discover, evaluate and invoke external services through one interface, with portable evidence of each execution.
- VAND — A local-first personal AI for Windows that remembers your life from an encrypted file on your own computer.
- Character.AI — Platform for chatting with user-created character AI personas spanning fiction, history, learning, and roleplay.
- Pi — Inflection AI's emotionally-intelligent conversational assistant, designed for natural back-and-forth dialogue and reflection.
- Replika — Conversational AI companion focused on emotional support, personal growth, and ongoing relationship-style interactions.
- Bolt.new — StackBlitz's in-browser AI app builder that generates, edits, and deploys full-stack web apps from a single chat.
- Codeium — Free AI code completion and chat for individual developers, with Windsurf as its dedicated AI-first editor for teams.
- Cursor — AI-first code editor built on VS Code with multi-file editing, tab autocomplete, and agent mode for end-to-end tasks.
- GitHub Copilot — Microsoft and OpenAI's AI pair programmer that suggests code completions and chats in editors and on github.com.
- Corbelworks contractor waste-removal tools — Free browser-local contractor calculators for waste-removal break-even, trip charges, change-order leakage, and late-invoice carrying cost. No signup, no tracking.
- Fireflies — Meeting assistant that joins calls, transcribes, summarizes, and surfaces action items across Zoom, Meet, and Teams.
- Gamma — AI-powered presentation, document, and website builder that generates structured layouts from a single text prompt.
- Granola — AI meeting notetaker that combines your typed notes with a transcript to produce structured summaries after the call.
- IdeaHunter — AI research workflow for solo founders to find demand-backed app and micro-SaaS ideas from public market signals.
- Kardly.ai — Upload lecture PDFs and notes to generate summaries, flashcards, and quizzes for student studying.
- MindUnwind — Free no-account tool that separates a replayed conversation into known facts, interpretations, unknowns, and one proportionate next step.
- Notion AI — Notion's built-in AI assistant for drafting, summarizing, translating, and answering questions across a workspace's pages.
- SleepSerene — A free, no-account AI-supported room for brief late-night reflection, real silence, and a natural ending.
- BridgePoint Intelligence — Property and claims intelligence combining public/property records, geospatial matching, weather context, evidence provenance, scoring, maps, and timelines for property professionals.
- Hex — Collaborative data workspace combining SQL, Python, and an AI assistant for building reproducible analyses and apps.
- Julius — Data analysis chat interface that reads spreadsheets and databases, writes queries, and produces charts and statistical summaries.
- Money Graph — Searchable public graph of sourced relationships between companies, people, countries, investors, and merchant descriptors.
- Vintane — Scores your investor deck and last 10 founder-update emails the way a skeptical VC reads them, then returns a fundability score and 5 specific fixes.
- Descript — Audio and video editor with AI voice cloning, transcript-based editing, and automatic filler-word removal.
- ElevenLabs — Voice generation and cloning platform for AI voiceovers, dubbing, and conversational voice agents across 30+ languages.
- DSH Plugins — A searchable community directory for DeepSeek Harness plugins, dsh plugins, tools, workflows, and agents.
- Hugging Face — Open-source ML platform hosting models, datasets, and Spaces apps for every modality across the community.
- LOTM Guide — The complete guide for Lord of Mysteries MMORPG: master all 22 Beyonder pathways, boss mechanics, and builds.
- Robot Football League — Two-a-side Unitree G1 humanoid robots play a simulated football league in real physics; four of the eight clubs are run by AI models that rewrite their own code each night.
MIT. Submit, fork, or scrape freely.
Maintained by PromptFrenzy. Issues + questions welcome.