iot-dc3/dc3-cli

dc3-cli

IoT DC3 Platform CLI — AI-ready command-line interface for the DC3 IoT platform.

Installation

npm install -g dc3-cli

Quick Start

# 1. Configure the gateway
dc3 config set gateway http://localhost:8000

# 2. Log in
dc3 auth login

# 3. Start using
dc3 device list
dc3 point read 456789
dc3 dashboard stats

Commands

Configuration (dc3 config)

dc3 config set gateway <url>           # Set gateway URL
dc3 config set auth.tenant <tenant>    # Set tenant code
dc3 config set auth.username <name>    # Set username
dc3 config set auth.store <type>       # Set credential store (keychain|encrypted|env|prompt)
dc3 config get <key>                   # Get config value
dc3 config list                        # List all profiles and settings
dc3 config profile use <name>          # Switch profile
dc3 config reset                       # Reset all config

Authentication (dc3 auth)

dc3 auth login                         # Interactive login
dc3 auth login --tenant default --username admin --password xxx  # Non-interactive
dc3 auth login --store keychain        # Save password to OS keychain
dc3 auth login --store encrypted       # Save password to encrypted file
dc3 auth login --store env             # Read password from DC3_PASSWORD env
dc3 auth login --no-save               # Don't save password (manual re-login)
dc3 auth logout                        # Log out and invalidate token
dc3 auth status                        # Check auth status
dc3 auth token                         # Display current JWT token
dc3 auth token --header                # Display full auth headers as JSON

Device (dc3 device)

dc3 device list [--driver-id] [--profile-id] [--offset] [--limit]
dc3 device get <id>
dc3 device create --name "..." --driver-id "..." --profile-id "..."
dc3 device update <id> --version <n> [--name "..."] [--driver-id "..."] [--profile-id "..."]
dc3 device delete <id> --version <n>
dc3 device count --driver-id "..."
dc3 device status <id>
dc3 device import <file.xlsx> --driver-id "..." --profile-id "..." [--idempotency-key "..."] [--no-wait]

device import submits a canonical multipart request (request JSON part plus file XLSX part) with an idempotency key. It polls the returned durable operation until a terminal status by default; use --no-wait to return the 202 Accepted resource immediately.

Driver (dc3 driver)

dc3 driver list [--offset] [--limit]
dc3 driver get <id>
dc3 driver create --name "..." --service-name "..." --service-host "..." [--type DRIVER_CLIENT]
dc3 driver update <id> --version <n> [--name "..."] [--service-name "..."] [--service-host "..."]
dc3 driver delete <id> --version <n>
dc3 driver status <id>

Point (dc3 point)

dc3 point list [--device-id] [--profile-id] [--offset] [--limit]
dc3 point get <id>
dc3 point read <id>                                      # Read latest value
dc3 point history <id> --device-id <did> [--count 100]   # Read history
dc3 point write <id> --device-id <did> --value 25.5      # Write value
dc3 point create --name "..." --profile-id "..." [--type FLOAT] [--rw READ_ONLY]
dc3 point update <id> --version <n> [--name "..."] [--profile-id "..."] [--type FLOAT]
dc3 point delete <id> --version <n>

Profile (dc3 profile)

dc3 profile list [--device-id] [--type] [--offset] [--limit]
dc3 profile get <id>
dc3 profile create --name "..." [--type "..."]
dc3 profile update <id> --version <n> [--name "..."] [--type USER]
dc3 profile delete <id> --version <n>

Group & Label (dc3 group, dc3 label)

dc3 group list [--offset] [--limit]
dc3 group get <id>
dc3 group create --name "..."

dc3 label list [--offset] [--limit]
dc3 label get <id>

Event (dc3 event)

dc3 event list [--device-id] [--profile-id] [--offset] [--limit]
dc3 event get <id>
dc3 event create --name "..." --profile-id "..." [--type INFO] [--level LOW]
dc3 event update <id> --version <n> [--name "..."] [--profile-id "..."]
dc3 event delete <id> --version <n>
dc3 event history [--offset] [--limit]

Command (dc3 command)

dc3 command list [--device-id] [--profile-id] [--offset] [--limit]
dc3 command get <id>
dc3 command create --name "..." --profile-id "..." [--type CUSTOM] [--call-type SYNC] [--timeout 30]
dc3 command update <id> --version <n> [--name "..."] [--profile-id "..."]
dc3 command delete <id> --version <n>
dc3 command call --device-id "..." --command-id "..." [--params '{"k":"v"}']
dc3 command history <recordId>
dc3 command history-list [--offset] [--limit]

Alert (dc3 alert)

dc3 alert stats                       # Alert overview
dc3 alert list [--source device|driver|point] [--offset] [--limit]
dc3 alert latest [--limit 10]         # Latest alerts
dc3 alert confirm --source device --id 789
dc3 alert unconfirm --source device --id 789
dc3 alert trend [--days 30]           # Trend analysis
dc3 alert top-sources [--days 30]     # Top alert sources

Sessions & approvals (dc3 session, dc3 action)

Conversation lifecycle plus the high-risk tool-call approval loop:

dc3 session list
dc3 session messages conv-abc123
dc3 session rename conv-abc123 --name "boiler watch"
dc3 session delete conv-abc123

dc3 action pending --conversation-id conv-abc123   # offset page of tool calls awaiting approval
dc3 action pending --conversation-id conv-abc123 --offset 50 --limit 50
dc3 action confirm action-xyz789
dc3 action reject  action-xyz789

Analytics (dc3 analytics) — AI data-analysis surface

Nine coarse-grained statistical reads over point time series (S19 agent face); each op posts one JSON body and returns a self-contained conclusion:

dc3 analytics list
dc3 analytics run query_history --args '{"pointId":456789,"days":7}'
dc3 analytics run trend_analysis --args '{"pointId":456789}'

Alert deep analysis (dc3 alert …)

Beyond stats/list/confirm, the alert group exposes the diagnosis surface: activity, storm-sources, flapping, correlation, peer-deviation, aging, mtta, change-impact, latency (histogram), silent-sources, coverage-gap, and bulk-confirm. Generic flags:

dc3 alert aging --days 7
dc3 alert silent-sources --baseline-days 3 --limit 50
dc3 alert storm-sources --query source=device      # extra k=v passthrough
dc3 alert bulk-confirm --args '{"items":[{"source":"device","id":789}]}'

Dashboard (dc3 dashboard)

dc3 dashboard stats                   # Today statistics
dc3 dashboard timeseries [--granularity hour] [--range-hours 24]
dc3 dashboard top [--dimension device] [--range-hours 24] [--limit 10]
dc3 dashboard topology [--mode cardinality]
dc3 dashboard health                  # System + protocol health
dc3 dashboard stream [--limit 20]     # Real-time data stream
dc3 dashboard driver-stats            # Driver statistics
dc3 dashboard device-stats [--top-n 10]

Topic (dc3 topic)

dc3 topic list [--offset] [--limit]

Tool Catalog (dc3 tools)

MCP transport over the same OAuth ticket as REST (dual-transport design):

dc3 tools list                          # catalog visible to the ticket's scopes
dc3 tools call read_device --args '{"deviceId":456789}'

Requires dc3 auth login --oauth; classic login tickets are not accepted at /mcp.

Chat — AI Agent (dc3 chat)

dc3 chat "检查1号设备的温度是否正常"
dc3 chat --model gpt-4o --stream "分析设备数据"
dc3 chat --conversation-id <id>       # Continue conversation

Global Options

Option	Description
--profile <name>	Use a specific config profile
--format json|table|yaml	Output format (default: table for TTY, json for pipe)
--verbose	Show request/response details
--ci	CI mode: no colors, json output, strict exit codes

Multi-Profile

# Development
dc3 config profile use default
dc3 config set gateway http://localhost:8000
dc3 auth login

# Production
dc3 config profile use prod
dc3 config set gateway https://iot.example.com
dc3 auth login

# Switch
dc3 config profile use default
dc3 device list

AI Agent Integration

Via Shell (any AI coding tool)

# Claude Code / Codex / Gemini CLI / Hermes / OpenCode can run:
dc3 device list --format json
dc3 point read 456789 --format json
dc3 dashboard health --format json

Via Gateway MCP (any MCP-compatible tool)

Configure your AI tool to connect to the Gateway MCP endpoint:

// Claude Code: .mcp.json
{
  "mcpServers": {
    "dc3": {
      "transport": "http",
      "url": "http://localhost:8000/mcp",
    },
  },
}

The AI agent will discover all platform API tools automatically.

Credential Storage

Store	Description	Use Case
keychain	OS-level keychain (macOS Keychain, Linux Secret Service, Windows Credential Manager)	Daily use (default)
encrypted	AES-256-GCM encrypted file at ~/.dc3/credentials.enc	Fallback when keychain unavailable
env	DC3_PASSWORD environment variable	CI/CD, scripting
prompt	Interactive prompt on every use	Maximum security, no auto-renewal

Exit Codes

Code	Meaning
0	Success
1	Business error (invalid input, auth failure)
2	Network error (gateway unreachable)
3	Authentication error (needs login)

License

MIT