"Six months ago, everyone was talking about MCPs. And I was like, screw MCPs. Every MCP would be better as a CLI."

— Peter Steinberger, Founder of OpenClaw Watch on YouTube (~2:39:00) | Lex Fridman Podcast #491

Tisane CLI

A production-ready command-line interface for the Tisane NLP and content moderation API. Analyze text, detect language, moderate content, extract entities, and analyze sentiment directly from your terminal.

Disclaimer: This is an unofficial CLI tool and is not affiliated with, endorsed by, or supported by Tisane Labs.

Features

• Parse — Full NLP analysis pipeline with entity and topic extraction
• Moderation — Detect abuse, hate speech, and policy violations
• Sentiment — Analyze sentiment scores and labels
• Entities — Extract named entities (people, places, organizations)
• Transform — Grammar correction and text transformation
• JSON output — All commands support --json for scripting and piping
• Colorized output — Clean, readable terminal output with chalk

Why CLI > MCP

MCP servers are complex, stateful, and require a running server process. A CLI is:

• Simpler — Just a binary you call directly
• Composable — Pipe output to jq, grep, awk, and other tools
• Scriptable — Use in shell scripts, CI/CD pipelines, cron jobs
• Debuggable — See exactly what's happening with --json flag
• AI-friendly — AI agents can call CLIs just as easily as MCPs, with less overhead

Installation

npm install -g @ktmcp-cli/tisane

Authentication Setup

Get your API key from the Tisane developer portal.

Configure the CLI

tisane config set --api-key YOUR_API_KEY

Commands

Configuration

tisane config set --api-key <key>
tisane config show

Parse (Full NLP Analysis)

# Full NLP analysis
tisane parse text --text "Apple is looking at buying U.K. startup for $1 billion"

# With specific language
tisane parse text --text "Bonjour le monde" --language fr

# Detect language
tisane parse detect --text "こんにちは世界"

Content Moderation

# Check for abusive content
tisane moderation check --text "Some text to moderate"

# With language specification
tisane moderation check --text "Texto para moderar" --language es

Sentiment Analysis

# Analyze sentiment
tisane sentiment analyze --text "I love this product, it's amazing!"
tisane sentiment analyze --text "This is terrible, never buying again." --language en

Entity Extraction

# Extract named entities
tisane entities extract --text "Elon Musk founded SpaceX in 2002 in Hawthorne, California"

# Extract topics
tisane entities topics --text "The Federal Reserve raised interest rates amid inflation concerns"

Text Transformation

# Fix spelling and grammar
tisane transform fix --text "Ths sentance has sum errers in it"

# With language
tisane transform fix --text "Il y a des erreurs" --language fr

JSON Output

All commands support --json for machine-readable output:

# Get full analysis as JSON
tisane parse text --text "Hello world" --json

# Extract only entities
tisane entities extract --text "..." --json | jq '.[] | {type, text}'

# Get sentiment score
tisane sentiment analyze --text "..." --json | jq '.sentiment'

Examples

Content moderation pipeline

# Read text from file and moderate
cat user_comment.txt | xargs -I{} tisane moderation check --text "{}"

# Batch moderate with jq
echo '["Hello", "Hate message here"]' | jq -r '.[]' | while read line; do
  tisane moderation check --text "$line" --json
done

Entity extraction workflow

# Extract all person entities
tisane entities extract --text "..." --json | jq '.[] | select(.type == "Person") | .text'

Contributing

Issues and pull requests are welcome at github.com/ktmcp-cli/tisane.

License

MIT — see LICENSE for details.

Part of the KTMCP CLI project — replacing MCPs with simple, composable CLIs.