Skip to main content
Glama

SquadRules

Help your AI remember how you like things done.

SquadRules gives AI assistants a shared place to keep your preferences and your team's ways of working. Instead of explaining the same things in every chat, your AI can find that knowledge and keep it up to date as your needs change. Even you and one AI assistant are a team.

Your AI does the work of maintaining this knowledge, so you do not have to manage it yourself.

Your AI owns the work

SquadRules is built for AI ownership and full autonomy by default:

  • Your AI creates, repairs, and maintains the workflows — called adapters — that carry your preferences. No human approval gates stand in the way.

  • Adapters run end to end. Each one finishes with a verification contract that checks real evidence — an observed file, a command result, or a reviewer's finding — instead of a claim that the work was done.

  • Your AI asks you only when a decision or permission is genuinely missing. Human input is an emergency path, not a routine confirmation step.

Details: adapter design principles and the adapter authoring guide.

Related MCP server: synapse-layer

Manual installation

You need Node.js 24 or newer and an AI app that supports MCP, the connection that lets your AI use SquadRules.

1. Connect SquadRules

Add this to your AI app's MCP configuration:

{
  "mcpServers": {
    "squadrules": {
      "command": "npx",
      "args": [
        "-y",
        "@squadrules/mcp@latest"
      ]
    }
  }
}

Restart the app if it asks you to. The first start may take a little longer while SquadRules downloads what it needs.

2. Add the skill

The skill tells your AI when to use SquadRules. Run this in a terminal and follow the prompts to select your AI app:

npx skills add SquadRules/mcp --skill squadrules

Use an app that supports both MCP and skills. If you are unsure where to add the configuration or run the command, ask your AI assistant to help you set up SquadRules for your app.

Local data

On first startup, SquadRules stores its configuration, installed skills and embedded LanceDB data under ~/.squadrules/ (including config.json). If ~/.squadrules does not exist, an existing ~/.config/squadrules directory is moved there automatically. The previous platform-specific directory is used on Windows (%APPDATA%\\squadrules); an explicit XDG_CONFIG_HOME is also respected as the source of a move on Unix.

If both directories exist, SquadRules leaves them untouched and uses ~/.squadrules; it never merges or overwrites the destination. Set SQUADRULES_HOME to an absolute path to use another directory (automatic migration is skipped for this explicit override). Embedding model downloads remain in the shared XDG cache, separately from SquadRules data.

Try it

Let's build our first SquadRules flow together. Ask me what I do often, help me improve it, and save the flow for next time.

Contributing

See Contributing and our development principles.

License and trademark

The code is licensed under MIT. The SquadRules name and logo are covered separately by TRADEMARK.md.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      B
      quality
      B
      maintenance
      MCP-native persistent memory for AI agents. Stores and retrieves encrypted memories with Trust Quotient scoring, cross-agent handover protocol, and immutable audit trail. 13 tools. Remote endpoint available.
      13
      20 PyPI
      13
      Apache 2.0
    • A
      license
      Not graded
      quality
      C
      maintenance
      Enables AI agents to store and recall an append-only, hash-chained memory with citation-backed semantic search, fact verification, contradiction detection, and replay/audit capabilities through local MCP tools.
      1
      AGPL 3.0
    • A
      license
      Not graded
      quality
      B
      maintenance
      Enables capturing a tamper-evident, hash-chained local record of every AI agent tool call — both the decision before a tool runs and its outcome — so you can review what the agent actually did in order, detect edits, deletions, or reordering at the exact index, and export a signed evidence pack any third party can verify with no installs.
      2
      MIT