Bring your own Agent
Install @getfrontline/cli, generate personal API keys, and connect external AI agents or terminal workflows to Frontline.
Use the CLI to work with Frontline from your terminal.
Use it when you need repeatable terminal work: list agents, inspect workflows, query objects, switch profiles, get JSON output, or use Max in a developer context.
Check which workspace behavior, access rule, credential, usage signal, or data-model surface the control changes.
What Bring your own Agent is for
Bring your own Agent lets operators use Frontline with external AI agents and developer tools. The screen shows supported agent contexts, CLI installation, authentication, personal API keys, and Create Personal API Key.
This is the bridge between the product and external agent environments such as Claude, Cursor, Codex, ChatGPT, Windsurf, and similar tools.
Frontline CLI package
The public CLI package is @getfrontline/cli on npm: https://www.npmjs.com/package/@getfrontline/cli/. Install it with npm install -g @getfrontline/cli, or run npx @getfrontline/cli --help to inspect commands without a global install.
The package ships two binaries: frontline for Public API work such as agents, workflows, billing, tables, objects, and JSON-friendly scripting; and max for Max chat/admin terminal workflows. It requires Node 18 or newer.
CLI use cases
Use frontline for repeatable scripts: list agents, inspect workflows, query objects, validate configuration, or pipe JSON output into another tool.
Use max when you need Max from the terminal or want to connect an external agent such as Claude, Cursor, Codex, or ChatGPT with Frontline context.
What each control changes
Install Frontline CLI gives the terminal install command. Authenticate gives the login command and expects a personal API key. Create Personal API Key generates a key for agent access. Copy controls help move commands or keys into the operator's environment.
Personal API keys should be treated as credentials. Create them for specific external agent usage, rotate them if exposed, and delete keys that are no longer needed.
Operational outcome
External AI agents can work with Frontline context without asking operators to copy product data manually. That makes implementation faster while keeping access explicit and revocable.
Operational playbook
Use this page to identify what the control changes, which source context it uses, who owns the result, and where to verify it.
Use the page to identify the trigger, source data, owner, visible result, and review point before changing production behavior.
Best practices
Start with the operational job before changing configuration. Name the owner, define the trigger or source context, and decide how the result should be reviewed.
Start with one workflow, message, record update, or Max task before expanding. The team should know what changed, who owns it, and how to pause or adjust it.
Troubleshooting
If the result is wrong, debug in this order: source data, permissions, connected tools, required fields, workflow logs, then any generated output used later.
When debugging, open the related CRM record, Max activity, workflow run, or integration log first. Fix the source data or rule before adding more automation.
FAQs
What does Bring your own Agent control?
Generate personal API keys and connect external AI agents to Frontline through the CLI.
Where is the CLI package?
The public package is @getfrontline/cli on npm: https://www.npmjs.com/package/@getfrontline/cli/.
Which binaries does it install?
It installs frontline for Public API workflows and max for Max chat/admin terminal workflows.
Who should use this page?
Workspace admins, implementation owners, and operators responsible for configuring access, account behavior, developer integrations, or the data model.
How should this page fit into onboarding?
Use it to understand the product surface, inspect real UI states, and connect the concept to daily operating workflows before configuring production behavior.
What should I verify before using this in production?
Verify ownership, permissions, source context, failure behavior, and the handoff path so teammates can trust what the system does next.