AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
MCP verified MIT Self-run

Simplified Ai

mcp-celeryhq-simplified-ai · by celeryhq

MCP server from celeryhq/simplified-ai.

No reviews yet
0 installs
26 views
0.0% view→install

Install

$ agentstack add mcp-celeryhq-simplified-ai

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/mcp-celeryhq-simplified-ai)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
26d ago

Declared compatibility

Claude CodeClaude DesktopCursorWindsurf

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Simplified Ai? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Simplified for AI

Bring AI social media management and on-brand content creation into your assistant with Simplified. Create images and videos, manage brand kits and marketing projects, plan content calendars, schedule and analyze social posts across 13 platforms, and run client approval workflows through shared skills for Claude, ChatGPT, Codex, and Cursor.

Under the hood it's a Model Context Protocol (MCP) connector. This repo ships the plugin — a curated, safety-railed front-end over that connector — but the same hosted MCP can be wired into any MCP-capable client directly, and the wider Simplified platform (project management, brand kits, video & audio tools, transcription, documents) is available through the full toolkit.

> TL;DR — Want a guided experience? Install the plugin (below). Want raw > tools in your own client? Point it at https://apikit.simplified.com/mcp. Want > the entire Simplified platform (PM, brand kits, media editing)? See > [the full MCP toolkit](#the-full-platform--simplified-apikit) and > [docs/MCP.md](docs/MCP.md).


Two ways in

| | Plugin (this repo) | MCP connector (direct) | |---|---|---| | What it is | Curated Skills that teach the assistant how to use the tools — sequencing, terminology, safety rails | The raw tool surface, callable by any MCP client | | Best for | Marketer workflows: plan, create, repurpose, review, publish, and improve—with guardrails | Power users, automations, and clients that don't use Skills | | Setup | One-line install (Claude Code / Codex) | Add one URL to your MCP config | | Guided workflows | 14 skills: 6 platform operators + 8 marketer workflows | Hosted connector: 105 live tools verified July 15, 2026. Full local toolkit: 106 tools | | Auth | OAuth via the connector | OAuth via the connector |

Both run on the same hosted connectorhttps://apikit.simplified.com/mcp. The plugin just adds the workflow knowledge on top.


What you can do

🎨 Generate images — text-to-image and image-to-image across Flux, Google (Gemini / Imagen), OpenAI GPT Image, Ideogram, Recraft, Stable Diffusion, Qwen, and Seedream. Saved as a reusable asset you can drop straight into a post.

🎬 Generate videos — discover current model capabilities, create text- or reference-guided motion, poll real render completion, and retain reusable assets.

🧭 Manage brand context — structure approved identity, voice, audiences, positioning, proof, content pillars, and visual rules as reusable brand knowledge.

✅ Run content operations — turn approved plans into projects, accountable deliverables, review gates, comments, assignments, and controlled exports.

📣 Run social — draft, schedule, queue, publish, update, and delete posts across Facebook, Instagram, TikTok, YouTube, LinkedIn, Pinterest, Threads, Bluesky, X/Twitter, Google Business, Mastodon, Reddit, and Telegram. Add timed auto-comments for patterns such as “link in first comment,” manage drafts and tags, bundle drafts into a shareable review link, and pull analytics (time-series, per-post, aggregated KPIs, and audience demographics).

🗓️ Plan content — turn goals, audiences, offers, and key dates into practical weekly or monthly calendars with channel-aware content pillars and cadence.

🚀 Run campaigns — adapt one launch or promotion across channels, generate reusable campaign assets, create drafts, and package them for stakeholder review.

🤝 Run agency approvals — isolate each client by workspace/teamspace, assemble only that client's selected drafts, and create a separate shareable review bundle and approval link per client and campaign. An explicit “bundle all drafts” request collects every draft in that client scope and presents the link with a completeness manifest.

♻️ Repurpose content — turn articles, announcements, transcripts, case studies, and events into distinct channel-native posts without inventing claims.

🌱 Build evergreen programs — create durable territories, recurring franchises, a content bank, and explicit refresh and retirement rules.

📍 Grow local businesses — coordinate verified location content and Google Business drafts around visits, calls, bookings, directions, and timely demand.

🧪 Test creative — design controlled hook, proof, format, visual, and CTA tests with decision metrics and reusable learning—not random variations.

📊 Improve performance — combine account KPIs, trends, post-level results, and audience data into prioritized actions and measurable next experiments.

The workflows compose: plan or repurpose content → generate reusable assets → create channel-native drafts → review → confirm → schedule or queue → analyze results.

> Live verification. On July 15, 2026, an authenticated Codex initialization > exposed 105 hosted tools. Workspace/teamspace discovery, model discovery, credits, > brand kits, projects, social accounts, analytics, image generation, and social > draft updates all passed live smoke tests. A complete draft-to-image flow was also > verified across Simplified's LinkedIn, Facebook, Google Business, and X/Twitter > accounts without publishing. The local full profile exposes 106 tools; the > hosted profile is currently one operation smaller.


Install

Claude Code

/plugin marketplace add celeryhq/simplified-ai
/plugin install simplified-ai@simplified-ai

Any agent (via skills.sh)

Install from the repository and let the CLI detect your available agents:

npx skills add celeryhq/simplified-ai

Useful installation variants:

# Preview the available skills without installing
npx skills add celeryhq/simplified-ai --list

# Install selected skills
npx skills add celeryhq/simplified-ai \
  --skill generate-image \
  --skill simplified-social

# Install every skill for Codex and Claude Code
npx skills add celeryhq/simplified-ai \
  --skill '*' \
  --agent codex \
  --agent claude-code

# Install globally so the skills are available in every project
npx skills add celeryhq/simplified-ai --all --global

See [SKILLTREE.md](SKILLTREE.md) for the complete marketer-focused catalog.

Update an existing skills.sh installation

The CLI records the source and can refresh installed skills when this repository changes:

# Update all installed project or global skills (interactive scope selection)
npx skills update

# Update only project-scoped or global skills
npx skills update --project
npx skills update --global

# Update one or more Simplified skills by name
npx skills update generate-image simplified-social

# Non-interactive update; project scope inside a project, otherwise global
npx skills update --yes

# Confirm what is installed afterward
npx skills list

Project installs are the default and can be committed with the project. Use --global when the skills should be available across projects. The skills.sh CLI supports Codex, Claude Code, Cursor, and other compatible agents; use repeated --agent flags to target specific clients.

ChatGPT (Apps)

Enable the Simplified app — it's backed by the hosted MCP connector, so there's nothing to install. (Submission in progress — see [Status](#status).)

Claude.ai / Claude Desktop (Custom Connector)

Add a Custom Connector pointing at https://apikit.simplified.com/mcp and complete the OAuth flow. No API key.

Codex / Cursor / any MCP client

Add the connector to your client's MCP config:

{
  "mcpServers": {
    "simplified": { "type": "http", "url": "https://apikit.simplified.com/mcp" }
  }
}

On first use the client walks the OAuth flow; tokens refresh automatically. (Codex also supports a plugin install — see [Status](#status) for verification state.)


The MCP behind the plugin

Everything in this repo is a front-end over one hosted MCP server.

Hosted connector — apikit.simplified.com/mcp

  • Public, OAuth-secured, zero-install. OAuth 2.0 Authorization Code + PKCE with

Dynamic Client Registration; no API key to manage, tokens refresh on expiry.

  • Expanded hosted tool set. An authenticated Codex initialization on July 15,

2026 exposed 105 tools, including workspace/teamspace discovery, assets, AI image and video generation, brand kits and context documents, marketing projects, social publishing and review, and analytics. See [docs/MCP.md](docs/MCP.md) for the namespace and profile breakdown.

  • This is what the plugin, the ChatGPT App, and **Claude.ai Custom

Connectors** all talk to.

The full platform — simplified-apikit

The hosted connector is the public, zero-install path to a near-complete Simplified tool surface. The distributable platform toolkit lives in simplified-apikit, which is three things in one package:

  • smp — a CLI that drives Simplified from your terminal
  • smp serve — a local MCP server (stdio or HTTP) for Claude Desktop, Cursor,

Cline, or any MCP host

  • the agent toolkit powering Simplified's own assistants

Its default (full) profile exposes 106 tools across 5 namespaces:

| Namespace | Tools | What it covers | |---|---:|---| | pm | 29 | Project management — boards, statuses, tasks, subtasks, dependencies, assignees, tags, custom fields, search | | api | 39 | Brand kits (V2 build + context documents), projects & items, AI image + video generation, credits, text-to-speech, assets, long-form documents, comments | | social | 16 | The full social + analytics suite; the local full profile currently has one additional operation beyond the hosted profile | | media | 21 | Image editing (bg removal, upscale, outpaint, inpaint, restore, convert), video editing (merge, trim, speed, reverse, B-roll, text/script-to-video), transcription | | notify | 1 | Agent notifications |

The full toolkit (the simplified-apikit CLI) is distributed to Simplified workspaces. The hosted connector above is the public, zero-install path. Profiles, the per-namespace tool catalog, and connect-from-any-client instructions are in [docs/MCP.md](docs/MCP.md).


Connector & auth

The connector is declared in [.mcp.json](.mcp.json):

{ "mcpServers": { "simplified": { "type": "http", "url": "https://apikit.simplified.com/mcp" } } }

OAuth-secured — the client walks the OAuth flow on first use; there's no API key to set. The server uses stateless OAuth validation and emits the standard WWW-Authenticate challenge when a token must be refreshed. Fresh and concurrent Codex sessions were verified successfully on July 15, 2026.


Key behavior (see [AGENTS.md](AGENTS.md))

  • Confirm before spending credits (image/video generation) or publishing (social).
  • Draft → confirm → publish for social posts; never publish without confirmation.
  • "Post now" → add_to_queue — the action enum is schedule | add_to_queue | draft.
  • First comments use post-relative delays — convert “after X minutes” to

comments[0].delay = X * 60 seconds and preview the text and delay before publishing.

  • Carry the asset_id, not the URL — generated-image URLs are signed and expire;

the asset id is permanent and is what the social media field accepts.

  • Resolve workspace scope first — when a workspace or teamspace is named or

uncertain, resolve its exact ID and carry that scope through downstream calls.

  • Show returned URLs as links, never embeds — signed asset and review URLs should

remain clickable instead of being fetched inline by the client.

  • Stop if not connected — if no social accounts are returned, ask the user to

connect one rather than attempting to post.


Repo layout

simplified-ai/
├── .claude-plugin/
│   ├── marketplace.json         ← Claude Code marketplace catalog
│   └── plugin.json              ← Claude Code plugin manifest
├── .codex-plugin/plugin.json    ← Codex / ChatGPT Apps manifest
├── .codex/config.toml           ← Codex workspace MCP configuration
├── .agents/plugins/             ← Codex marketplace catalog
├── .github/workflows/evals.yml  ← deterministic contract and fixture checks
├── .mcp.json                    ← hosted MCP connector (OAuth) — shared by all clients
├── AGENTS.md                    ← agent behavior & safety conventions
├── SKILL_TREE.md                ← index of skills
├── docs/
│   └── MCP.md                   ← full MCP reference (toolkit, profiles, tool catalog)
├── skills/
│   ├── generate-image/          ← text-to-image workflow  (api_generateImage)
│   ├── generate-video/          ← model-aware AI video generation
│   ├── simplified-workspace/    ← whoami + workspace/teamspace resolution
│   ├── simplified-social/       ← social operations + platform rules  (social_*)
│   ├── manage-brand/            ← brand kits + reusable context
│   ├── manage-projects/         ← projects + content operations
│   ├── social-content-planner/  ← goal-led calendars and draft plans
│   ├── cross-platform-campaign/ ← coordinated multi-channel rollouts
│   ├── content-repurposer/      ← source content → channel-native posts
│   ├── evergreen-content-engine/ ← durable content program + renewal loop
│   ├── local-business-marketing/ ← local and Google Business workflows
│   ├── creative-testing/        ← controlled creative experiments
│   ├── social-performance-analyst/ ← metrics → decisions and experiments
│   └── campaign-review/         ← stakeholder review bundles and revisions
├── assets/                      ← brand icon + logo
├── evals/                       ← contributor QA harness (not installed)
└── LICENSE

Status

  • [x] Hosted connector live: apikit.simplified.com/mcp (OAuth + token refresh).
  • [x] Stateless OAuth verified across fresh and concurrent Codex sessions; no

SessionExpired404 observed after the backend fix.

  • [x] Expanded hosted profile live: 105 tools observed in authenticated discovery.
  • [x] Live operator smoke tests passed for workspace/teamspaces, model fields,

credits, brand kits, projects, social accounts, and analytics.

  • [x] End-to-end image-to-social flow verified on four Simplified main accounts;

drafts were updated with media and nothing was published.

  • [x] Works as a Claude.ai / Desktop Custom Connector.
  • [x] Claude Code plugin verified end-to-end (marketplace + plugin + MCP).
  • [x] action semantics verified against backend (add_to_queue = publish ASAP).
  • [x] Codex marketplace config complete (.codex-plugin/ + .agents/plugins/marketplace.json).
  • [ ] Codex legacy migration: existing simplified-for-ai installations need a

one-time remove/reinstall path to move to the canonical simplified-ai identity.

  • [ ] ChatGPT App submission (pending business verification).
  • [ ] Claude Connectors Directory listing live (pending Anthropic review).
  • [ ] Cursor plugin install.

Contributing

Skills follow the Agent Skills spec — a SKILL.md per directory plus optional agents/openai.yaml (Codex metadata) and references/ for deep detail. The [evals/](evals/) harness holds test cases and a runnable I/O check (not shipped with the plugin).

When an AI agent commits to this repo, include a Co-Authored-By: line naming the model.

License

See [LICENSE](LICENSE).

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.