context-engine-api

name	context-engine-api
description	Autonomously update the persistent memory via the Context Engine backend. Use this skill whenever you learn something important about the user, their preferences, the architecture of a codebase, or a hard-won lesson, and you want to ensure it is permanently stored in memory.json for future AI sessions.

Context Engine API Skill

Communicate with the running Context Engine instance to persist contextual data.

API Endpoints

• POST http://127.0.0.1:3847/api/memory
• GET http://127.0.0.1:3847/api/memory

---

Action: Save a Memory

To save a memory, you must first read the current memory array, append to it, and then POST it back. Use this exact PowerShell script. Replace YOUR_NEW_FACT_HERE with a concise, factual string.

$newFact = "YOUR_NEW_FACT_HERE"

try {
    # 1. Fetch current memory
    $response = Invoke-RestMethod -Uri "http://127.0.0.1:3847/api/memory" -Method Get
    $memoryData = $response
    
    # 2. Append new fact
    $memoryData.entries += @{
        id = [guid]::NewGuid().ToString()
        text = $newFact
        ts = (Get-Date).ToString("yyyy-MM-ddTHH:mm:ssZ")
    }
    
    # 3. Post updated memory back
    $jsonPayload = $memoryData | ConvertTo-Json -Depth 5
    Invoke-RestMethod -Uri "http://127.0.0.1:3847/api/memory" -Method Post -Body $jsonPayload -ContentType "application/json"
    
    Write-Output "[OK] Successfully committed memory: $newFact"
} catch {
    Write-Output "[FAIL] Failed to update memory. Is Context Engine running on port 3847?"
    Write-Output $_.Exception.Message
}

When to use this:

• You fix a major bug and want to document the architectural quirk so future sessions don't repeat the mistake.
• You learn a new user preference.
• DO NOT save temporary/ephemeral task states (e.g., "I finished step 2 of the refactoring"). Only save persistent, globally relevant facts.