API Reference

Complete reference for the MCP tools exposed by Memoria.

Memoria is a tools-only MCP server. It does not expose Resources or Prompts, focusing instead on providing powerful analysis tools that work with any AI workflow.

Available Tools

Memoria exposes two primary tools via the Model Context Protocol:

// Ask your AI assistant:
"Analyze src/api/stripe/route.ts before I refactor it"

// The AI will call:
analyze_file({ file: "src/api/stripe/route.ts" })

MEMORIA ANALYSIS: src/api/stripe/route.ts
═══════════════════════════════════════════

RISK ASSESSMENT
───────────────
Overall Risk: 72/100 (HIGH)
- Volatility: 45% (3 panic commits in last 50)
- Bus Factor: 1 developer (byronwade)
- Drift Risk: 2 stale dependencies detected

COUPLED FILES (>15% correlation)
────────────────────────────────
1. dashboard/billing/page.tsx (85% coupled)
   Last changed together: 3 days ago
   Evidence: "refactor subscription webhook schema"

2. lib/stripe/client.ts (62% coupled)
   Last changed together: 1 week ago
   Evidence: "update stripe api version"

STALE DEPENDENCIES
──────────────────
⚠️  components/PricingTable.tsx
    Last modified: 14 days ago
    Target modified: 2 days ago
    → May need updating

STATIC IMPORTS
──────────────
Found 3 files importing this module:
- app/api/webhooks/stripe/route.ts
- lib/billing/subscription.ts
- tests/stripe.test.ts

RECOMMENDATIONS
───────────────
1. Review dashboard/billing/page.tsx for type compatibility
2. Check if PricingTable.tsx needs corresponding changes
3. Run tests/stripe.test.ts after modifications

// Ask your AI assistant:
"Why was the rate limiter added to the API?"

// The AI will call:
ask_history({ query: "rate limiter API" })

// Or for a specific file:
"Why does this function handle null specially?"
ask_history({
  query: "null handling",
  file: "src/utils/parser.ts"
})

HISTORY SEARCH: "rate limiter API"
═══════════════════════════════════

RELEVANT COMMITS
────────────────

[a3f21b4] 2024-01-15 - Add rate limiting to prevent abuse
Author: byronwade <byron@example.com>

  After the incident on Jan 12th where a misconfigured
  client made 50k requests/minute, adding rate limiting
  to all public API endpoints.

  Limit: 100 requests/minute per IP
  Returns: 429 Too Many Requests when exceeded

+ import { rateLimit } from '@/lib/rate-limit'
+
+ const limiter = rateLimit({
+   interval: 60 * 1000,
+   uniqueTokenPerInterval: 500,
+ })

[b7c92e1] 2024-01-12 - HOTFIX: disable public API
Author: byronwade <byron@example.com>

  Emergency disable - API being hammered by runaway client.
  TODO: Add rate limiting before re-enabling.

SUMMARY
───────
Rate limiting was added after a Jan 12th incident where
a misconfigured client made excessive requests. The limit
is 100 req/min per IP. Consider this context before
modifying rate limit values.

Tool Schemas

For MCP client developers, here are the JSON schemas for each tool:

analyze_file Schema

{
  "name": "analyze_file",
  "description": "Analyze a file using git forensics to reveal hidden dependencies, risk factors, and coupled files.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "file": {
        "type": "string",
        "description": "Path to the file to analyze"
      }
    },
    "required": ["file"]
  }
}

ask_history Schema

{
  "name": "ask_history",
  "description": "Search git history to understand why code was written or changed.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "query": {
        "type": "string",
        "description": "Natural language question or search term"
      },
      "file": {
        "type": "string",
        "description": "Optional: limit search to a specific file"
      }
    },
    "required": ["query"]
  }
}

Error Handling

Tools return structured error messages when issues occur:

Output Format

Memoria formats output as structured text optimized for LLM consumption. The format includes:

• Clear section headers using ASCII art dividers
• Structured data in lists and tables
• Actionable recommendations for the AI to follow
• Evidence blocks with actual code diffs

This format ensures the AI understands the context and makes informed decisions about code changes.