AI-powered Adobe After Effects automation via MCP. Chat in a CEP panel or connect external agents, with 16 tools for editing, previews, recovery, reusable scripts, and multi-instance workflows.
README
English | 简体中文
<details open> <summary>Automatic setup: paste this prompt into the MCP client you want to use</summary>Install and configure ae-mcp for the current client receiving and executing
this prompt. Do not default to Claude Code or configure another client.
Before changing anything, inspect the local ae-mcp extension and native plug-in
files and versions, plus this client's existing MCP registration and actual
connection, read-only. Existence alone does not prove readiness. Compare with
the latest official stable release assets available for my platform:
- Client registration already correct: keep it; install only missing or damaged files.
- Files complete and versions suitable: only add or repair client registration.
- Both ready and current: verify without reinstalling.
- Newer version available: upgrade existing components, preserving user settings
and other MCP configuration; do not downgrade. If versions cannot be established
or platform assets are unavailable, explain rather than blindly overwrite.
Use the release ZXP and platform-native plug-in for needed file installation or
upgrade, then open Window > Extensions > ae-mcp in After Effects.
Identify this client's configuration method and location, preserving all other
MCP entries. Use http://127.0.0.1:11488/mcp if it supports Streamable HTTP;
for stdio only, configure `npx -y ae-mcp-jkdg` using this client's format.
If automatic editing is unavailable, give precise manual steps for this client.
Ask which client only if the target truly cannot be identified; never silently
fall back to Claude Code. Keep the AE panel open, refresh/reconnect or start a
new session as this client requires, then call ae_status to verify. If user
action is required first, state that verification is still pending.
</details>
Let AI read your After Effects project, build animations, inspect the results, and save successful operations as reusable tools.
ae-mcp is an open-