Script and Automate AI Interactions
Turn repeated AI workflows into version-controlled, reusable scripts - Bash, Python, Node.js, TypeScript - executed on demand.
Why it matters
Transform complex AI interactions into reusable, version-controlled scripts. Automate repetitive tasks, ensure consistency, and enable team collaboration on AI-driven workflows.
Outcomes
What it gets done
Create and manage executable AI interaction scripts.
Automate complex tasks with a single command.
Organize and version control your automation library.
Share and reuse proven AI workflows across your team.
Install
Add it to your toolbox
Run in your project directory:
curl -fsSL https://spark.entire.vc/get/vb-scriptflow | bash Capabilities
Tools your agent gets
Add a new script with name, description, content, language, and tags
Update an existing script
View script details
Search and filter scripts by name, description, or tags
Execute a script with arguments
Delete a script
Overview
ScriptFlow MCP Server
ScriptFlow turns successful multi-step AI interactions into persistent, executable, version-controlled scripts (Bash, Python, JS, TS) that can be searched, edited, and run with arguments across sessions. Use it for tasks you expect to repeat, to save time and ensure consistent execution. Not useful for one-off tasks with no reuse value.
What it does
ScriptFlow is a script workflow management system built on MCP that converts successful, complex AI interactions into persistent, executable scripts - saved, searchable, and reusable across sessions instead of repeating the same long conversation every time.
When to use - and when NOT to
Use it when you've worked out a multi-step task with Claude and want to save it as a script you (or your team) can run instantly next time, with guaranteed consistent behavior instead of re-deriving the steps through conversation. It supports Bash, Python, JavaScript, and TypeScript scripts, each stored with metadata for searching by name, description, or tags. This is especially useful for saving tokens and time on repetitive tasks, and for sharing proven workflows across a team. Do not use it for one-off tasks that won't be repeated - the value comes from reuse and consistency across multiple runs.
Capabilities
script_add: add a new script with name, description, content, language, and tags.script_edit: update an existing script.script_get: view a script's details.script_list: search/filter scripts by name, description, or tags.script_run: execute a script with arguments.script_rm: remove a script.
Scripts are stored as {name}.{ext} plus a {name}.json metadata file, supporting .sh, .py, .js, and .ts.
How to install
Configure your MCP client (e.g. Claude Desktop):
{
"mcpServers": {
"scriptflow": {
"command": "npx",
"args": ["scriptflow-mcp"],
"env": { "SCRIPTFLOW_SCRIPTS_DIR": "/your/custom/path" }
}
}
}
Environment variables: SCRIPTFLOW_SCRIPTS_DIR (default /tmp/scriptflow-mcp/scripts), SCRIPTFLOW_TIMEOUT (execution timeout in ms, default 30000). Debug with the MCP Inspector: npx @modelcontextprotocol/inspector npx scriptflow-mcp.
Who it's for
Developers and teams who repeat complex, multi-step AI workflows and want to convert them into reliable, version-controlled scripts instead of re-explaining the task in conversation each time.
Source README
ScriptFlow MCP Server
ScriptFlow is a script workflow management system built on the Model Context Protocol (MCP). It transforms complex, repetitive AI interactions into persistent, executable scripts that can be managed, version-controlled, and reused across sessions.
Why ScriptFlow?
- Transform Workflows into Scripts: Convert successful AI interactions into reusable, executable scripts
- Save Time & Tokens: Execute complex tasks instantly instead of repeating long conversations
- Guaranteed Consistency: Scripts run the same way every time, eliminating AI randomness
- Powerful Management: Search, organize, edit, and version your automation library
- Team Collaboration: Share proven workflows across your organization
Features
- Add/Edit/Remove scripts with metadata
- List/Search scripts by name, description, or tags
- Execute scripts with arguments
- Multi-language support (Bash, Python, Node.js, TypeScript)
Setup
Add to your MCP client config (e.g., Claude Desktop claude_desktop_config.json):
{
"mcpServers": {
"scriptflow": {
"command": "npx",
"args": ["scriptflow-mcp"],
"env": {
"SCRIPTFLOW_SCRIPTS_DIR": "/your/custom/path"
}
}
}
}
Environment Variables:
SCRIPTFLOW_SCRIPTS_DIR: Scripts directory (default:/tmp/scriptflow-mcp/scripts)SCRIPTFLOW_TIMEOUT: Execution timeout in ms (default:30000)
Available Tools
script_add- Add new script (name, description, content, language, tags)script_edit- Update existing scriptscript_get- View script detailsscript_list- Search/filter scriptsscript_run- Execute script with argumentsscript_rm- Remove script
Quick Example
// Add script
{"name": "hello", "description": "Hello script", "content": "echo 'Hello!'"}
// Run script
{"name": "hello", "args": ["World"]}
// List scripts
{"search": "hello"}
Storage
Scripts stored as {name}.{ext} + {name}.json metadata files.
Supports: Bash (.sh), Python (.py), JavaScript (.js), TypeScript (.ts)
Testing
# Run tests
npm test
# Test with MCP Inspector
npx @modelcontextprotocol/inspector npx scriptflow-mcp
FAQ
Common questions
Discussion
Questions & comments · 0
Sign In Sign in to leave a comment.