MCP Integration

MCP (Model Context Protocol) is the standard for connecting AI assistants to external tools. Instructionly uses MCP to deliver your instructions to AI-powered IDEs.

Getting Your MCP Credentials

Go to Settings > MCP Keys
Click Create MCP Key
Enter a name for the key
Select an expiration period (1 day, 7 days, 15 days, 30 days, or No expiry)
Click Create
Copy your secret immediately — it's only shown once

Your MCP Key consists of:

MCP Key ID — The public identifier
MCP Secret — The private key (shown only once)
Expiration — When the key will stop working (optional)

Key expiration

For security, consider setting an expiration on your keys. Expired keys are automatically rejected and you'll need to create a new one.

Using MCP Keys

Add Instructionly to your AI tool's MCP configuration. First, install the CLI, then add this configuration:

json

{
  "mcpServers": {
    "instructionly": {
      "command": "instructionly",
      "env": {
        "MCP_KEY_ID": "YOUR_KEY_ID",
        "MCP_SECRET": "YOUR_SECRET",
        "MCP_WORKSPACE": "YOUR_WORKSPACE_SLUG"
      }
    }
  }
}

Replace the environment variables with your credentials:

MCP_KEY_ID — Your MCP Key ID
MCP_SECRET — Your MCP Secret
MCP_WORKSPACE — Your workspace slug

MCP keys are user-scoped

MCP keys are tied to your user account, not individual workspaces. Use the same key with different workspace values to access multiple workspaces.

Revoking MCP Keys

If a key is compromised:

Go to Settings > MCP Keys
Find the key to revoke
Click Revoke
Confirm the action

The key will immediately stop working.

Using ion

Once configured, you can activate Instructionly in two ways:

Option 1: Simply mention "ion" in your prompt — the AI will automatically use Instructionly:

prompt

ion how should I implement authentication in this project?

Option 2: Use the built-in MCP prompt command to explicitly invoke Instructionly:

prompt

/instructionly/ion how should I implement authentication in this project?

Instructionly will:

Analyze your request
Load relevant skills from your workspace
Apply matching instructions
Return AI-ready guidance

Available MCP Tools

When connected, your AI assistant has access to these Instructionly tools:

Tool	Description
`ion`	Delivers Instructionly workflow instructions to your AI assistant
`inspector`	Analyzes your request to determine which skills and actions apply
`directives`	Retrieves actionable instructions based on detected skills
`restore_context`	Restores session context after conversation compaction
`wrap_up`	Reviews work for instruction compliance before finalizing
`check_balance`	Shows your current credit balance
`check_usage`	Displays recent credit consumption history
`list_skills`	Shows available skills in your workspace
`get_help`	Provides documentation and support resources
`service_status`	Checks Instructionly platform health
`workspace_info`	Shows workspace details and statistics
`notify_checkpoint`	Sends notification when task completes or needs user input
`set_checkpoint_notification`	Updates notification preference mid-session

How Instructions Are Applied

When you use ion, the following process occurs:

Request Analysis — The inspector tool analyzes your prompt to identify the action type (code, test, review, etc.)
Skill Matching — Relevant skills are matched based on the detected action
Instruction Loading — Instructions from matched skills are loaded
Context Injection — Instructions are provided to your AI assistant as context
Credit Consumption — Credits are deducted based on usage