fix: MCP humanize tool rendering findings as [object Object]

[iVibudh]

fix: MCP humanize tool returning [object Object] for all findings

What's broken

The humanize tool in the MCP server renders every finding as [object Object].
Any agent or LLM client calling this tool gets no usable information about what to fix.

Root cause

humanize() returns critical and important as arrays of objects:

{
  pattern: "AI vocabulary",
  text: "delve",
  line: 1,
  suggestion: "Tier 1 AI word: \"delve\". Use a simpler, more specific alternative.",
  confidence: "high"
}

The MCP handler was iterating them as if they were strings:

for (const s of suggestions.critical) {
  output += `- ${s}\n`; // → "- [object Object]"
}

Fix

Format each object into a readable line with line number, pattern name, flagged text, and suggestion:

output += `- L${s.line}: [${s.pattern}] "${s.text}" → ${s.suggestion}\n`;

Before / after

Before:

### 🔴 Critical (Dead giveaways)
- [object Object]
- [object Object]
- [object Object]

After:

### 🔴 Critical (Dead giveaways)
- L1: [AI vocabulary] "delve" → Tier 1 AI word: "delve". Use a simpler, more specific alternative.
- L1: [AI vocabulary] "synergies" → Tier 1 AI word: "synergies". Use a simpler, more specific alternative.
- L1: [Chatbot artifacts] "I hope this helps" → (remove)

Files changed

• mcp-server/index.js — fixed critical and important loop formatting in the humanize case