by sipyourdrink-ltd · sipyourdrink-ltd/bernstein
Deterministic orchestrator across 40+ CLI agents — auditable, reproducible runs that don't depend on a single model.
Bernstein gives you a deterministic execution model on top of CLI agents (Claude Code, Codex, Gemini CLI, Aider, etc.). Same prompt + same inputs = same audited trail. Useful for compliance and CI use.
%APPDATA%\Claude\claude_desktop_config.json{
"mcpServers": {
"bernstein-skill": {
"command": "git",
"args": [
"clone",
"https://github.com/sipyourdrink-ltd/bernstein"
],
"_inferred": true
}
}
}Open Claude Desktop → Settings → Developer → Edit Config. Restart after saving.
{
"mcpServers": {
"bernstein-skill": {
"command": "git",
"args": [
"clone",
"https://github.com/sipyourdrink-ltd/bernstein"
],
"_inferred": true
}
}
}Cursor uses the same mcpServers schema as Claude Desktop. Project config wins over global.
{
"mcpServers": {
"bernstein-skill": {
"command": "git",
"args": [
"clone",
"https://github.com/sipyourdrink-ltd/bernstein"
],
"_inferred": true
}
}
}Click the MCP Servers icon in the Cline sidebar, then "Edit Configuration".
{
"mcpServers": {
"bernstein-skill": {
"command": "git",
"args": [
"clone",
"https://github.com/sipyourdrink-ltd/bernstein"
],
"_inferred": true
}
}
}Same shape as Claude Desktop. Restart Windsurf to pick up changes.
{
"mcpServers": [
{
"name": "bernstein-skill",
"command": "git",
"args": [
"clone",
"https://github.com/sipyourdrink-ltd/bernstein"
]
}
]
}Continue uses an array of server objects rather than a map.
{
"context_servers": {
"bernstein-skill": {
"command": {
"path": "git",
"args": [
"clone",
"https://github.com/sipyourdrink-ltd/bernstein"
]
}
}
}
}Add to context_servers. Zed hot-reloads on save.
claude mcp add bernstein-skill -- git clone https://github.com/sipyourdrink-ltd/bernsteinOne-liner. Verify with claude mcp list. Remove with claude mcp remove.
When to use: Compliance asks for a record of what was sent, by whom, with what result.
Outcome: Reproducible runs with full traceability.
Audit + guardrails together
| Tool | Inputs | When to call | Cost |
|---|---|---|---|
| run / replay / export | (see docs) | Three flows | 1 call |
rotate audit log without retention policyNon-determinismSet seed / temperature where the backend supports it
| Alternative | When to use it instead | Tradeoff |
|---|---|---|
| Single-agent CLI | You don't need audit | No reproducibility |