DoubleX AgentesDocs

CLI › Commands

Agents

List, inspect, create, edit and delete agents from the terminal.

Terminal
doublex agents list | get | create | update | delete

List#

Terminal
doublex agents list
doublex agents list --mine --search sales --limit 10
OptionWhat it's for
--search <text>Filters by name and description.
--mineOnly your agents and the ones shared with you. For admins, without this option the list includes every agent in the organization.
--limit <n>Shows only the first n.
--jsonReturns { "agents": [...] } with every field.

The table shows the ID, the name, the model (provider/model), the scope, the number of conversations and whether you can edit the agent.

View an agent#

Terminal
doublex agents get agt_8f3a2b1c9d0e

Shows the model, the reasoning effort, the scope, the description, how many skills, connectors, Brain documents and subagents it uses, and the full instructions. With --json, it also returns the agent's recent conversations and triggers.

Create#

Terminal
doublex agents create \
  --name "CRM Analyst" \
  --description "Answers questions about the sales pipeline" \
  --instructions-file instrucoes.md \
  --emoji 📊 \
  --model anthropic/claude-sonnet-5-5 \
  --effort high

The CLI shows the new agent's ID. The agent starts in the Personal scope; to share it with a team or the organization, use the agent's page in the app.

OptionWhat it's for
--name <name>Required. Up to 60 characters.
--description <text>Up to 300 characters.
--instructions <text>The agent's instructions (up to 20,000 characters). Use - to read from stdin.
--instructions-file <file>Reads the instructions from a file.
--emoji <emoji>The agent's emoji.
--model <provider/model>Model. Use an ID from doublex models list, or auto for DoubleX Auto, which picks the model for each message. Without it, the agent uses the organization's default model.
--effort low|medium|highReasoning effort. The default is medium.
--args-json <json> / --args-file <file>Other fields of the create body, as JSON (for example, color and avatar). The options above take priority.

To create an agent from a ready-made template, send the template key in --args-json:

Terminal
doublex agents create --args-json '{"templateKey": "briefing-diario"}'

Edit#

Terminal
doublex agents update agt_8f3a2b1c9d0e --model auto --effort low
doublex agents update agt_8f3a2b1c9d0e --instructions-file instrucoes.md

Accepts --name, --description, --instructions, --instructions-file, --model, --effort and --args-json/--args-file. Only what you pass is changed.

With --args-json you can change any field of the agents API, for example the skills, connectors and Brain documents the agent uses:

Terminal
doublex agents update agt_8f3a2b1c9d0e --args-json '{
  "skillIds": ["skl_3e8f1a2b4c5d"],
  "connectorIds": ["con_7a9b1c3d5e2f"],
  "knowledgeIds": ["kno_2c4e6a8b0d1f"]
}'

You can only edit agents you have edit permission on (the CAN EDIT column in the list).

Delete#

Terminal
doublex agents delete agt_8f3a2b1c9d0e --yes

Without --yes, the CLI only tells you what would be deleted. Deleting an agent also removes its conversations, triggers and the files on its computer.