CLI User Guide¶

The cbintel CLI is an interactive terminal interface for exploring and experimenting with the graph execution pipeline.

Installation¶

cd cli
npm install

Quick Start¶

# Interactive mode
npm run dev

# Run a test query
npm run dev -- -t "Research AI safety"

# List available templates
npm run dev -- --templates

# Run auto experiments
npm run dev -- --auto 5

Running Modes¶

Interactive Mode¶

npm run dev

Launches a full-screen terminal UI with: - Status bar showing current template, mode, and workspace - Message history with syntax highlighting - Input box for queries and commands

Test Mode¶

npm run dev -- -t "query" [--template name]

Runs a single query non-interactively. Useful for scripting and CI/CD.

Examples:

npm run dev -- -t "Research AI safety"
npm run dev -- -t "Elon Musk" --template person_deep_research
npm run dev -- -t "Apple Inc" --template corporate_intelligence
npm run dev -- -t "https://cnn.com" --template news_50_states

Auto Mode¶

npm run dev -- --auto [count]

Runs automated experiments to discover pipeline capabilities and edge cases.

Examples:

npm run dev -- --auto        # Run 5 experiments (default)
npm run dev -- --auto 10     # Run 10 experiments
npm run dev -- -a 3          # Short form, 3 experiments

Production Mode¶

npm run build    # Compile TypeScript
npm run start    # Run compiled code

Commands Reference¶

Command	Description
`/help`	Show help panel with commands and templates
`/templates`	Jump to templates list
`/clear`	Clear message history
`/exit`	Exit the CLI

Query Execution¶

Command	Description
`<query>`	Run query with current template
`/run <template> <input>`	Run specific template with input
`/use <template>`	Set default template for queries

Workspace Management¶

Command	Description
`/workspace create <name>`	Create new workspace
`/workspace list`	List all workspaces
`/workspace use <id>`	Set active workspace
`/workspace info`	Show current workspace details
`/workspace clear`	Clear active workspace

Auto Mode¶

Command	Description
`/auto`	Toggle auto-experimentation mode
`/auto stats`	Show experiment statistics
`/auto stop`	Stop auto mode

Templates¶

Research Templates¶

Template	Description	Parameters
`simple_search`	Basic web search and summarization	`query`
`person_deep_research`	Comprehensive background research on any individual	`person_name`, `context?`
`opponent_deep_dive`	Multi-source research on a political opponent	`opponent_name`, `geo_focus?`
`corporate_intelligence`	Deep research into company operations and leadership	`company_name`, `ticker?`, `industry?`

Geographic Templates¶

Template	Description	Parameters
`news_50_states`	Screenshot a news site from all 50 US states	`url`
`geo_news_comparison`	Compare how news appears in different regions	`url`, `regions`

Change Tracking Templates¶

Template	Description	Parameters
`recent_website_changes`	Track what changed on a website recently	`url`, `days_back?`
`historical_change_tracker`	Track website changes over time with archive analysis	`target_urls`, `date_from?`, `date_to?`
`position_tracker`	Track stance/position changes over time	`person_name`, `topic`

Analysis Templates¶

Template	Description	Parameters
`competitor_comparison`	Side-by-side analysis of multiple companies	`companies[]`
`topic_sentiment_tracker`	Analyze sentiment around a topic across sources	`topic`, `subtopics?`
`media_coverage_analysis`	Analyze media coverage patterns and bias	`subject`

Advanced Templates¶

Template	Description	Parameters
`autonomous_research_agent`	AI-driven research with iterative discovery	`query`, `depth?`
`tor_search_crawl`	Search and crawl dark web sources	`query`

Usage Examples¶

Basic Research¶

# Start interactive mode
npm run dev

# Type a query
> Research recent developments in quantum computing

# Or use a specific template
> /run person_deep_research "Sam Altman"

Workspace Workflow¶

# Create a workspace for a project
> /workspace create "Q1 Competitor Analysis"

# Run queries - jobs are automatically associated
> /run corporate_intelligence "OpenAI"
> /run corporate_intelligence "Anthropic"
> /run corporate_intelligence "Google DeepMind"

# Check workspace stats
> /workspace info

Template Switching¶

# Set default template
> /use topic_sentiment_tracker

# Now all queries use this template
> AI regulation
> climate policy
> healthcare reform

# Switch to another template
> /use person_deep_research
> Elon Musk
> Jensen Huang

Auto Experimentation¶

# Interactive auto mode
> /auto
# Experiments run automatically...
> /auto stats    # Check progress
> /auto stop     # Stop experiments

# Command-line auto mode
npm run dev -- --auto 20

Auto Mode Details¶

Auto mode runs experiments to discover pipeline capabilities and test edge cases.

What Gets Tested¶

Category	Examples
Topic Queries	AI safety, quantum computing, campaign finance, CRISPR trials
Person Research	Elon Musk, Sam Altman, Jensen Huang
Company Research	OpenAI, Anthropic, Google DeepMind
Edge Cases	Empty queries, unicode, emojis, XSS attempts, SQL injection

Statistics Tracked¶

Total experiments run
Success/failure count and rate
Average execution duration
Recent experiment history

Sample Output¶

🤖 cbintel-cli AUTO MODE

Running 5 experiments...

======================================================================

[1/5] topic:tech
  Template: simple_search
  Query: "AI safety regulations 2024"
  Status: ...... ✓ 5.4s (4 stages, 6 ops)

[2/5] company
  Template: corporate_intelligence
  Query: "OpenAI"
  Status: ................... ✓ 20.4s (5 stages, 18 ops)

[3/5] edge:unicode/Japanese
  Template: simple_search
  Query: "日本語テスト"
  Status: .... ✓ 3.2s (4 stages, 6 ops)

======================================================================

📊 SUMMARY

  Total:     5
  Succeeded: 5 (100.0%)
  Failed:    0 (0.0%)
  Avg time:  12.3s

Configuration¶

Environment Variables¶

Variable	Default	Description
`CBINTEL_API_URL`	`http://127.0.0.1:9003`	Jobs API endpoint

Example¶

# Use production API
CBINTEL_API_URL=http://intel.nominate.ai:9003 npm run dev

# Use local development server
CBINTEL_API_URL=http://localhost:9003 npm run dev

Command-Line Flags¶

Flag	Short	Description
`--help`	`-h`	Show help message
`--test <query>`	`-t`	Run non-interactive test
`--template <name>`		Template for test mode
`--auto [count]`	`-a`	Run auto experiments
`--count <n>`	`-n`	Number of experiments
`--templates`	`-T`	List available templates

Architecture¶

┌─────────────────────────────────────────────────────┐
│  cbintel [simple_search]              [human] /help │
├─────────────────────────────────────────────────────┤
│                                                     │
│  You: Research AI safety                            │
│  Agent: Running template: simple_search...          │
│  Agent: Completed: 4 stages, 6 operations           │
│                                                     │
├─────────────────────────────────────────────────────┤
│  > _                                                │
└─────────────────────────────────────────────────────┘
         │
         ▼
┌─────────────────────────────────────────────────────┐
│              Jobs API (port 9003)                   │
│  POST /api/v1/jobs/graph → Job submission           │
│  GET  /api/v1/jobs/{id}  → Status polling           │
└─────────────────────────────────────────────────────┘
         │
         ▼
┌─────────────────────────────────────────────────────┐
│              Graph Executor                         │
│  Stages → Operations → Results                      │
└─────────────────────────────────────────────────────┘

Troubleshooting¶

"Raw mode is not supported"¶

This error occurs when running interactive mode without a TTY (e.g., in a pipe or CI). Use test mode instead:

npm run dev -- -t "query"

API Connection Failed¶

Check that the Jobs API is running:

curl http://127.0.0.1:9003/

Or specify the correct URL:

CBINTEL_API_URL=http://intel.nominate.ai:9003 npm run dev

Template Not Found¶

List available templates:

npm run dev -- --templates

CLI User Guide¶

Installation¶

Quick Start¶

Running Modes¶

Interactive Mode¶

Test Mode¶

Auto Mode¶

Production Mode¶

Commands Reference¶

Navigation¶

Query Execution¶

Workspace Management¶

Auto Mode¶

Templates¶

Research Templates¶

Geographic Templates¶

Change Tracking Templates¶

Analysis Templates¶

Advanced Templates¶

Usage Examples¶

Basic Research¶

Workspace Workflow¶

Template Switching¶

Auto Experimentation¶

Auto Mode Details¶

What Gets Tested¶

Statistics Tracked¶

Sample Output¶

Configuration¶

Environment Variables¶

Example¶

Command-Line Flags¶

Architecture¶

Troubleshooting¶

"Raw mode is not supported"¶

API Connection Failed¶

Template Not Found¶

See Also¶