mcp-serverSAFE
MCP server for the Sync API — give AI agents the ability to create lipsync videos, manage assets, and more
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
An open-source Model Context Protocol (MCP) server for the Sync API. Gives AI agents the ability to create lipsync videos, manage assets, check generation status, and more.
Tools are auto-generated from the Sync OpenAPI spec at startup. As new API endpoints ship, they become available to agents automatically — no server update needed.
Supported Clients
Quick Start
Claude Web (claude.ai)
No installation required — connect directly from your browser.
- Go to claude.ai → Settings → Integrations
- Click Add custom connector
- Enter Name:
Syncand URL:https://mcp.sync.so/mcp - Click Add, then Connect
- Log in with your Sync account when prompted
No API key needed — authentication is handled via OAuth.
Claude Code
claude mcp add sync -- npx -y @sync.so/mcp-server --api-key YOUR_API_KEY
Or add to .mcp.json in your project root:
{
"mcpServers": {
"sync": {
"command": "npx",
"args": ["-y", "@sync.so/mcp-server"],
"env": {
"SYNC_API_KEY": "your-api-key"
}
}
}
}Claude Desktop
Add to claude_desktop_config.json:
{
"mcpServers": {
"sync": {
"command": "npx",
8d4ba2a5983cOBSERVED · 2026-10-11Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add mcp-server --env OAUTH_REGISTRATION_SECRET=${OAUTH_REGISTRATION_SECRET} --env OPENAI_APPS_CHALLENGE_TOKEN=${OPENAI_APPS_CHALLENGE_TOKEN} --env SYNC_API_KEY=${SYNC_API_KEY} -- npx -y @sync.so/[email protected]{
"mcpServers": {
"mcp-server": {
"command": "npx",
"args": [
"-y",
"@sync.so/[email protected]"
],
"env": {
"OAUTH_REGISTRATION_SECRET": "${OAUTH_REGISTRATION_SECRET}",
"OPENAI_APPS_CHALLENGE_TOKEN": "${OPENAI_APPS_CHALLENGE_TOKEN}",
"SYNC_API_KEY": "${SYNC_API_KEY}"
}
}
}
}Exposed tools (16)
4 read · 12 write · 0 destructive.
| Tool | Risk | Description |
|---|---|---|
assets_create-upload-url | write | |
create-dialogue-preview | write | Preview timed word changes/removals in the original speaker’s voice. This is a synthesis action. Use a completed transcription, or a prior preview of this source when reopening. Confirm edits first. Never retry an ambiguous creation automatically. This is not arbitrary full-transcript replacement. |
create-dialogue-transcription | write | Transcribe a saved single-speaker video for timed-word editing. Transcription is unbilled and identical sources reuse their existing job. Does not replace dialogue or generate a video. |
create-dialogue-video | write | Generate the final video from an approved completed audio preview matching the current edits. Confirm the estimate first. Reuse the same idempotency key and inputs for recovery; poll generate_get-generation for the returned ID. |
create-lipsync | write | Create a lipsync video with exactly one visual input (image or video) and one driver (audio or script). |
create-translate-and-dub | write | Translate a video’s speech and lip-sync it in the target language. Requires a saved video asset, selected project, model, and persistent idempotency key. |
estimate-dialogue-video | read | Estimate final video cost using the completed dialogue preview and canonical server retiming. Does not generate or reserve credits. |
get-dialogue-options | write | Check whether the connected API supports timed-word Edit Dialogue. Does not transcribe, synthesize or generate. |
get-dialogue-preview | write | Read one existing Edit Dialogue audio preview. Partial completion must be disclosed, not described as every edit succeeding. |
get-dialogue-transcription | read | Read the same transcription job until complete. Requires its original saved video and accessible project. |
get-translation-options | read | Read the source and target languages supported by the current translation API. Does not create a generation or estimate pricing. |
hold | read | |
open-sync-app | write | Open Sync to browse existing projects, choose media and configure a video generation. Requesting the interface does not start a generation or spend credits. The result only confirms the request; do not claim the interface rendered unless the user confirms it. |
open-upload-widget | write | Display the Sync image/audio uploader in ChatGPT. The widget can stage a selected local image or audio file as a Sync asset and start a lipsync generation. It does not accept video files. |
projects_get-generations | write | Read generation history for an accessible Sync project. Requires a projectId returned by projects_get-all. Pass limit and the last returned generation id as cursor to load another page. These are the same records used by the Sync web app. |
upload-media | write | Stage an image, video, or audio file in Sync asset storage and return a durable assetId. |
Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | PASS |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (3 observation(s))
- Network
- none-observed
- Shell
- none-observed
- Dependencies
- not all pinned
- Secrets in source
- none-found
Findings (15)
'http://169.254.169.254/private',
{ results: [{ address: '::ffff:169.254.169.254', family: 6 }] },http.request(`http://127.0.0.1:${upstreamAddress.port}/object`, options, callback),const origin = `http://127.0.0.1:${listener.address().port}`;http.request(`http://127.0.0.1:${upstreamAddress.port}/object`, options, callback as never),const origin = `http://127.0.0.1:${(listener.address() as AddressInfo).port}`;await run(`http://127.0.0.1:${(server.address() as AddressInfo).port}`, () => reached);@modelcontextprotocol/sdk, cors, express, express-rate-limit, ipaddr.js, undici, zod, @biomejs/biome
The HTTP server writes newline-delimited JSON diagnostics to stderr. Lifecycle events distinguish graceful pod termination from abrupt process loss, request logs include latency and abort state, and `
[Session E's previous exact-head report](https://linear.app/sync-labs/issue/CRAFT-6168#comment-5b306d82-3e81-4f2f-9e72-1d9e75e9ab19)
[Session C's reviewer kit](https://linear.app/sync-labs/document/muse-connector-launch-reviewer-kit-and-release-evidence-499e8a3c25d7)
deploy/chatgpt/0f4f454753a88e6264cf1393500c9d9dac8673040459642407230a0e1ddbb258/app.html.gz
deploy/chatgpt/5c28555d1843dfb74841ff1af3f5fff5cf8204f4ce52a8f6dabccbfc02d9c861/app.html.gz
deploy/chatgpt/a39e27ca488992ab54f148417646a041c7e6d667055dbc0fa5bcd674af76532b/app.html.gz
deploy/chatgpt/c048cd48c1d3bde0c9fafb1dd3cdd9ccc6dd45085060a92ed653b7cfce7592dd/app.html.gz
Gates applied: no_behavioural_pass.
8d4ba2a5983cfull audit observations/trust-audit/mcp-server/synchronicity-labs__mcp-server.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-11 | 8d4ba2a5983c | SAFE | B | 89 | first audit |
Questions
What is the mcp-server MCP server?
MCP server for the Sync API — give AI agents the ability to create lipsync videos, manage assets, and more
What tools does mcp-server expose?
16 in total: 4 read-only, 12 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.
Is mcp-server safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B.
What credentials does mcp-server need?
It reads OAUTH_REGISTRATION_SECRET, OPENAI_APPS_CHALLENGE_TOKEN and SYNC_API_KEY from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does mcp-server run?
It speaks stdio and streamable-http, so it runs as a local process your client starts. It is published on npm as @sync.so/mcp-server at 0.1.4.
How current is this page?
The grade is for one exact copy of the source (8d4ba2a5983c), read on 2026-10-11. The repository is watched and re-audited when it changes.