CommunityAbandoned· 10mo★ 4.3by API Network

PiAPI MCP

Generates media via Midjourney, Flux, Kling, Hunyuan, Udio, and Trellis through a single PiAPI account.

View repo Homepage

Safety & Trust

Delv Safety Grade: C

Score 58/100 · assessed 2026-04-28

Maintainer45

Permissions75

Supply chain65

Transparency55

Incidents100

PiAPI MCP is a community-maintained aggregator that routes generative media requests to six commercial APIs (Midjourney, Flux, Kling, Hunyuan, Udio, Trellis) through a single PiAPI account. The maintainer 'API Network' appears to be a small commercial entity rather than an established vendor, with limited public track record. The server itself is scoped: it makes outbound API calls and returns media URLs, requiring only a PIAPI_KEY environment variable. No filesystem or shell access is requested. Supply chain is standard npm distribution but the package is new and lightly reviewed. Transparency is moderate: the repository is open source with basic documentation, but lacks detailed changelog or active issue discussion. The core risk is dependency on PiAPI as a middleman: you trust them with your API key, they handle credentials for six downstream services, and any PiAPI outage or policy change affects all models. No known security incidents, but the trust surface is broad.

Lethal Trifecta (prompt-injection exposure)

ONE OF THREE

Private dataNo

Reads secrets, credentials, private files

Untrusted inputNo

Ingests web pages, PRs, issues, emails

External commsYes

Can send data outbound

Image generation passthrough.

Green flags

Narrow scope: outbound API calls only, no filesystem or shell access
Standard npm distribution with versioned package
Open source repository with visible code for review
Single environment variable (PIAPI_KEY) reduces credential sprawl
No known security incidents or takedowns

Red flags

Middleman architecture: PiAPI holds keys to six downstream generative APIs
Maintainer 'API Network' has minimal public track record or verification
New package with limited community review or adoption signals
Dependency on PiAPI service availability for all six models
No evidence of security audit or credential handling transparency

Permissions requested

Outbound networkAccess secretsExternal LLM call

Assessed by Delv Editorial using public metadata. Grades are advisory and update as the ecosystem changes. They do not replace your own review of permissions and code before granting an agent access to sensitive systems.

Install

npx -y @smithery/cli install piapi-mcp-server --client claude

Env vars needed: PIAPI_KEY

Review

PiAPI MCP is a single-account gateway to six generative media APIs: Midjourney, Flux, Kling, Hunyuan, Udio, and Trellis. Instead of juggling separate keys and endpoints for image, video, and music generation, you authenticate once with PiAPI and call whichever model you need through Claude. The promise is simple: one MCP server, six creative engines. I'd reach for this when prototyping apps that need multiple media types. Say you're building a product demo generator that needs Midjourney stills, Kling video clips, and Udio background music. Normally you'd wire up three separate integrations. Here, you configure one server and let Claude route requests. The workflow is tidy: ask Claude to generate a hero image via Midjourney, then request a 5-second product spin in Kling, then a 30-second soundtrack in Udio. All from the same conversation. The trade-off is obvious: you're dependent on PiAPI's uptime and pricing. If one of the upstream providers changes their terms or if PiAPI's relay goes down, you're stuck. The server itself is thin middleware, so quirks come from the underlying APIs. Midjourney prompts still need the same craft, Kling still has the same output limits. This isn't a magic wand, it's a convenience layer. Configuration is minimal. You need a PiAPI key, which means signing up for their service and understanding their credit model. The MCP server doesn't document rate limits or cost per call, so you'll need to check PiAPI's own pricing. That's a gap: I want to know what a Midjourney generation costs in credits before I start a batch job. Who shouldn't bother: anyone who only needs one provider. If you're a Midjourney-only shop, just use their API directly. This makes sense when you genuinely need multi-modal output and want to avoid managing multiple subscriptions. It's a solid aggregator for teams that ship creative tools, less compelling for hobbyists who can afford the friction of separate accounts.

Verdict

Install this if you're building apps that need image, video, and music generation and you'd rather manage one account than six. Skip it if you only use one provider or if you want full control over API versioning and fallback logic. It's a sensible aggregator, not a miracle.

Good at

One authentication flow for six generative media APIs saves setup time.
Useful for prototyping multi-modal apps without juggling separate accounts.
Works with Claude Desktop, Claude Code, and Cursor out of the box.
Straightforward install via Smithery CLI with minimal configuration.

Watch out

You're locked into PiAPI's pricing and uptime, with no direct fallback to the underlying providers.
No documented rate limits or per-call costs in the MCP repo, so budget planning requires digging into PiAPI's own docs.
Thin middleware means you still need to understand each provider's prompt syntax and output constraints.
Community-maintained, so updates depend on a single author rather than official support.

Use cases

Midjourney image generation
Kling video creation
Udio music generation
multi-provider creative apps

Getting started

1. Run `npx -y @smithery/cli install piapi-mcp-server --client claude` to install the server. 2. Sign up at PiAPI and grab your API key from the dashboard. 3. Add the server to your Claude Desktop config with `PIAPI_KEY` set to your key. Restart Claude. 4. Ask Claude to generate an image via Midjourney or a video via Kling to confirm the connection works. 5. Check PiAPI's credit usage after your first few calls so you understand the cost model before scaling up.

Works with

Claude DesktopClaude CodeCursor