max-chat
Send messages to the Max AI assistant and manage conversations from the terminal. Use when the user wants to chat with Max, send a quick question, start a new conversation, or open an interactive chat
SKILL.mdskills/vendor/max-chat/SKILL.md
GitHubPrerequisites
- The
maxCLI must be installed (npm i -g @getfrontline/cli— provides bothfrontlineandmaxbinaries). - The user must have a per-user API key saved:
max auth login <api-key>(orfrontline auth loginon the same profile).
Commands
Quick message (shorthand)
max "Your question here"
max --new "Start a fresh conversation"
This is the fastest way to send a single message to Max. It uses the last active conversation by default, or --new to start a fresh one.
Calls the Public API: POST /public/v1/max/conversations/message (Bearer = user API key). Responses use { "ok": boolean, "data": ... }. Default stdout is one-line JSON; no spinner lines.
| Flag | Description |
|---|---|
--new |
Start a new conversation instead of continuing the last one |
--conversation <id> |
Send to a specific conversation ID |
--no-wait |
Do not poll for the assistant reply (fire and forget) |
--timeout <seconds> |
Max seconds to wait for assistant reply (default: 60) |
--json |
Print only the POST response JSON (no poll) |
--pretty |
Print assistant plain text from data instead of JSON |
--profile <name> |
Use a specific Max CLI profile |
--base-url <url> |
Override Public API root (…/public/v1) for this run |
--api-base-url <url> |
Deprecated alias for --base-url |
--api-key <key> |
Override per-user API key for this run |
--debug |
Show HTTP request/response diagnostics |
Example:
# Quick question using last conversation
max "What agents do I have?"
# Start a new conversation
max --new "Help me set up a new workflow"
# Send to a specific conversation
max --conversation 123 "Continue from here"
Send a message (explicit command)
max chat send <message...> [--new] [--conversation <id>] [--json] [--debug]
Same as the shorthand above but under the chat send subcommand. Default stdout is one-line JSON; --pretty for assistant plain text from data.
Interactive chat (REPL)
max chat repl [--profile <name>] [--debug] [--timeout <seconds>]
max chat
Opens an interactive readline session for back-and-forth conversation with Max.
| REPL Command | Description |
|---|---|
:help |
Show available REPL commands |
:new |
Start a new conversation on next message |
:conv <id> |
Switch to a specific conversation ID |
:exit |
Exit the REPL |
Ctrl+C |
Exit the REPL |
Example:
# Start interactive chat
max chat
# Start with a specific profile
max chat repl --profile work --debug
Conversations (Public API)
max conversations list [--pinned <bool>]
max conversations get <id>
max conversations update <id> --data '{"key":"value"}'
max conversations abort <id>
max conversations pin <id>
max conversations unpin <id>
# alias: max conv list
See max conversations --help for flags (--profile, --base-url, --api-key, --debug).
Conversation management
- Max automatically remembers the last conversation ID in your profile. Subsequent messages continue that conversation.
- Use
--newto explicitly start a fresh conversation. - Use
--conversation <id>to jump to a specific conversation. - In the REPL, use
:newand:conv <id>to manage conversations interactively.
Troubleshooting
- "No Max API key found": run
max auth login <api-key>orfrontline auth login <api-key>on the same profile. - "Message cannot be empty": provide a non-empty message string.
- Timeout waiting for reply: increase
--timeoutor use--no-waitand check the conversation later. - Use
--debugto see the full HTTP request URL, headers, and response status for diagnosing issues. - Use
--jsonfor machine-readable output that can be piped tojqor other tools.
See also
- Full reference: https://docs.getfrontline.ai/docs (concepts) and https://docs.getfrontline.ai/cli (command reference).
- App UI walkthroughs (channels, billing, settings): https://help.getfrontline.ai.