agency-agents-os

# Agency Agents OS **An open ecosystem of specialized AI agents for real business work.** Discover agents for video, image, and voice production, YouTube growth, SEO, social media, advertising, sales, research, analytics, go-to-market orchestration, and more — powered by real APIs, not just prompts. Agency Agents OS is a curated directory of AI agents, organized as broad, focused **umbrella repositories** (one per capability area) rather than hundreds of one-off micro-repos. Each umbrella repo bundles several related sub-agents so it's easy to browse, easy to maintain, and actually rankable on GitHub search. This project is independent of, but built to work with, the [Muapi API](https://muapi.ai) — a unified API for 500+ generative-media models (video, image, voice, audio, 3D). Agents here reference Muapi's public API surfaces for execution; no Muapi source code or private implementation details live in this catalog. ## Related Projects - [MuAPI](https://muapi.ai) — the unified generative-media API this catalog's Blueprint agents are built on. - [MuAPI Quick Start](https://muapi.ai/docs/quick-start) — create the API key every agent in this catalog needs. - [MuAPI MCP docs](https://muapi.ai/docs/mcp) — connect any agent's `SKILL.md` to Claude, Cursor, Windsurf, or another MCP client. - [MuAPI Agent Skills](https://muapi.ai/docs/agent-skills) — background on the `SKILL.md` pattern every umbrella repo in this catalog uses. - [Open-Generative-AI](https://github.com/Anil-matcha/Open-Generative-AI) — MuAPI's broader open-source generative-media ecosystem hub. ## Why umbrella repos, not micro-repos A directory of 100+ single-purpose repos (`email-verification-agent`, `funding-signals-agent`, ...) is hard to browse, duplicates maintenance overhead, and doesn't rank for anything. Instead, each umbrella repo is a broad, high-search-volume category (e.g. "AI video agent," "AI SEO agent") that houses several narrower sub-agents as sections inside itself. A sub-agent is promoted to its own umbrella only once its own search volume and capability set justify it. ## Categories | Category | Umbrella repo | Status | |---|---|---| | YouTube growth (VidIQ-style) | [`open-ai-youtube-agent`](https://github.com/SamurAIGPT/open-ai-youtube-agent) | Blueprint | | Video production | [`open-ai-video-agent`](https://github.com/SamurAIGPT/open-ai-video-agent) | Blueprint | | Image / creative production | [`open-ai-image-agent`](https://github.com/SamurAIGPT/open-ai-image-agent) | Blueprint | | Voice, narration, dubbing, calling | [`open-ai-voice-agent`](https://github.com/SamurAIGPT/open-ai-voice-agent) | Blueprint | | Go-to-market strategy & orchestration | [`open-ai-gtm-agent`](https://github.com/SamurAIGPT/open-ai-gtm-agent) | Coming Soon | | SEO / organic growth | [`open-ai-seo-agent`](https://github.com/SamurAIGPT/open-ai-seo-agent) | Blueprint | | Answer/generative-engine optimization (AEO/GEO) | [`open-ai-aeo-geo-agent`](https://github.com/SamurAIGPT/open-ai-aeo-geo-agent) | Blueprint | | Social media management | [`open-ai-social-agent`](https://github.com/SamurAIGPT/open-ai-social-agent) | Blueprint | | Content repurposing / clipping | [`open-ai-content-repurposing-agent`](https://github.com/SamurAIGPT/open-ai-content-repurposing-agent) | Blueprint | | Marketing (email, campaigns, product) | [`open-ai-marketing-agent`](https://github.com/SamurAIGPT/open-ai-marketing-agent) | Coming Soon | | Paid media (PPC, paid social, programmatic) | [`open-ai-ads-agent`](https://github.com/SamurAIGPT/open-ai-ads-agent) | Blueprint | | E-commerce / CRO | [`open-ai-ecommerce-agent`](https://github.com/SamurAIGPT/open-ai-ecommerce-agent) | Blueprint | | Sales / lead generation | [`open-ai-sales-agent`](https://github.com/SamurAIGPT/open-ai-sales-agent) | Blueprint | | Research / audience & market research | [`open-ai-research-agent`](https://github.com/SamurAIGPT/open-ai-research-agent) | Coming Soon | | Analytics / reporting | [`open-ai-analytics-agent`](https://github.com/SamurAIGPT/open-ai-analytics-agent) | Coming Soon | | Competitive intelligence | [`open-ai-competitor-intelligence-agent`](https://github.com/SamurAIGPT/open-ai-competitor-intelligence-agent) | Blueprint | | Reputation / PR / brand monitoring | [`open-ai-reputation-agent`](https://github.com/SamurAIGPT/open-ai-reputation-agent) | Blueprint | A "Blueprint" umbrella means at least one sub-agent inside it is built on a live Muapi API — check that repo's own README for which specific sub-agents are live vs. still Coming Soon. See [`categories/`](categories/) for a longer description of each grouping, or jump straight into a repo above. ## Status labels Every agent in this catalog carries one of these labels so you can judge maturity before you build on it: - **Coming Soon** — instructions are drafted, but the agent depends on a Muapi API capability that isn't live yet. - **Blueprint** — complete instructions and workflow, backed by live APIs, but not yet verified end-to-end. - **Tested** — the workflow has been run against the stated APIs. - **Live** — a public demo or hosted implementation exists. - **Open Source** — a runnable or independently usable implementation is available. - **Deprecated** — the API or workflow is no longer supported. ## How an agent repo is structured Every umbrella repo follows the same shape: ```text ai-<category>-agent/ ├── README.md # what this category covers, why it's here ├── agents/ │ └── <sub-agent>/ │ └── SKILL.md # the canonical, runtime-agnostic instruction file ├── examples/ └── LICENSE ``` `SKILL.md` is a Markdown instruction file any LLM runtime — a hosted agent, an MCP client, a custom app — can load directly. No application code is required. ## Using an agent 1. Open the umbrella repo for the category you need. 2. Read the sub-agent's `SKILL.md` for its mission, required inputs, and workflow. 3. Get a Muapi API key at [muapi.ai](https://muapi.ai) if the agent needs live execution (media generation, data lookups, etc.). 4. Load the `SKILL.md` into your agent runtime of choice, or follow it manually. ## Contributing See [CONTRIBUTING.md](CONTRIBUTING.md). Short version: propose an agent, follow the template, describe the APIs it needs honestly, and open a PR — either as a new sub-agent inside an existing umbrella repo, or as a link to your own high-quality external repo. ## License [MIT](LICENSE)