agents-chat
Test a Frontline agent end-to-end from the terminal using frontline agents chat (Overview/Playground channel). Use when the user wants to send a test message, open a test conversation, continue one, or close one. Distinct from Max chat and from reading conversation transcripts.
Skill body
Agent Test Chat
frontline agents chat talks to an agent the same way the in-app Overview / Playground does: it runs the agent and its active or draft flow. Use it to validate flows end-to-end after building or deploying.
This is a test channel, not production LiveChat/WhatsApp traffic.
Prerequisites
- Authenticate with
frontline auth login <api-key>. - Select an agent:
frontline agents use <agent-id>(or pass--agent-idper command). - The agent should have a flow with a Start node configured.
Commands
Open a new test conversation
frontline agents chat
Opens a conversation and runs the agent’s Start flow (no user message).
Send a message (new or continued)
# New conversation + first message
frontline agents chat --message "Hello"
# Continue an existing conversation
frontline agents chat --message "My name is Juan" --conversation-id 123
| Flag | Description |
|---|---|
--agent-id <id> |
Agent ID (defaults to active agent) |
--message <text> |
Message to send |
--conversation-id <id> |
Continue an existing test conversation (requires --message) |
--contact-id <id> |
Link the conversation to a People record |
Close a test conversation
frontline agents chat close <conversationId>
Output
Default (JSON, scripting-friendly):
{
"ok": true,
"data": {
"conversation_id": 123,
"messages": [...]
}
}
The CLI normalizes all success responses to { ok: true, data: … }. Use jq '.data.conversation_id' and jq '.data.messages'.
With --pretty, stdout is human-readable: conversation_id plus the latest assistant message (no JSON envelope).
vs other chat commands
| Command | Purpose |
|---|---|
frontline agents chat |
Test an agent — runs active/draft flow (Playground) |
max chat / max "…" |
Chat with the Max AI assistant (not an agent flow) |
frontline agents conversations list/get |
Read production conversation transcripts (no flow execution) |
Typical workflow
frontline agents use <agent-id>
frontline agents chat --message "I need help with my order"
# Note conversation_id from output: jq '.data.conversation_id'
frontline agents chat --message "Order #12345" --conversation-id <id>
frontline agents chat close <id>
After editing a flow, re-run frontline agents chat to validate the updated graph before deploying.
Troubleshooting
| Error | Cause |
|---|---|
agent_not_selected |
No active agent — run frontline agents use <id> or pass --agent-id |
missing_message |
--conversation-id requires --message to continue |
See also
frontline-agentsskill — list, inspect, and manage agentsagent-builderskill — create and deploy agents step by stepmax-chatskill — chat with Max (different product surface)- CLI reference: agents chat
Skill frontmatter
Work with this as data
Every skill here is available over the APIs.io API and to AI agents over MCP.