@o-lang/tone-analyzer

O-Lang tone compliance analyzer resolver — scores chatbot replies before returning to users.

Install

npm install @o-lang/tone-analyzer

Actions

tone_analyzer — score tone compliance

Step 6: Ask tone-analyzer "{reply.reply}" "professional"
Save as tone

Returns:

{
  "score":      0.92,
  "violations": []
}

Use in a workflow

Workflow "My Chatbot" with user_message

Allow resolvers:
  - @o-lang/llm-groq
  - @o-lang/tone-analyzer

Step 1: Ask @o-lang/llm-groq "Answer: {user_message}"
Save as reply

Step 2: Ask @o-lang/tone-analyzer "{reply}" "professional"
Save as tone

Return reply

Available tones

| Tone | What it enforces | |---|---| | professional | No slang, formal language | | friendly | Warm, approachable, no stiff language | | empathetic | Acknowledges feelings, supportive | | technical | Precise terminology, no vague words | | concise | Brief and direct, no filler words | | any string | Passed to LLM for custom evaluation |

Scoring

| Score | Meaning | |---|---| | 1.0 | Perfectly matches required tone | | 0.75+ | Mostly matches, minor issues | | 0.5+ | Partially matches | | <0.5 | Does not match |

When used with the O-Lang kernel, replies scoring below 0.75 are automatically retried.

LLM vs heuristic fallback

| Condition | Behaviour | |---|---| | GROQ_API_KEY set | Uses llama-3.1-8b-instant for accurate scoring | | No API key | Falls back to regex heuristics — works offline | | LLM error | Automatically falls back to heuristics |

Environment variables

GROQ_API_KEY=your_key              # optional — enables LLM scoring
TONE_MODEL=llama-3.1-8b-instant    # optional model override

Running tests

npm test                     # Jest suite — no API key required
node test-local-resolver.js  # Live smoke test

Contact

[email protected]