multica-cli

name	multica-cli
description	Manage Multica resources (issues, agents, autopilots, skills, workspaces) and the local agent runtime daemon via the `multica` CLI. Use when the user runs `multica`, mentions Multica issues/agents/autopilots/skills/runtimes/workspaces, or needs to script Multica workflows.

Multica CLI

Command-line interface for Multica. Manage issues, assign agents, control the local daemon, and script everything that the web UI can do.

Quick start

multica login                 # first-time auth (opens browser, saves PAT to ~/.multica/config.json)
multica auth status           # verify login + active workspace
multica daemon status         # daemon must be online for local-runtime agents

For scripting and parsing, always pass --output json — the default for list commands is table (human-readable).

Common workflows

Issues

multica issue list --status open --output json
multica issue get ABC-123                              # accepts issue key or UUID
multica issue create --title "..." --description "..." # decodes \n, \r, \t, \\
multica issue assign ABC-123 --to "agent-slug"         # triggers a task immediately
multica issue status ABC-123 --set done
multica issue comment add ABC-123 --body "..."
multica issue runs ABC-123                             # execution history
multica issue run-messages <task-id> --issue ABC-123   # follow an agent run's messages
multica issue rerun ABC-123                            # re-queue with current assignee

For multi-line descriptions or content with literal backslashes, prefer --description-stdin:

printf 'line 1\nline 2\n' | multica issue create --title "..." --description-stdin

Agents and skills

multica agent list --output json
multica agent create --name "..." --runtime-id <id> --model claude-sonnet-4-6
multica agent skills attach <slug> --skill <skill-slug>
multica skill import --url https://github.com/org/repo   # also: clawhub.ai, skills.sh

Prefer --model <id> over passing --model inside --custom-args — some runtimes (codex app-server, openclaw) reject it there.

For secrets in --custom-env, use --custom-env-stdin or --custom-env-file (suggest mode 0600). Inline values leak to shell history and ps.

Autopilots (scheduled / triggered automations)

multica autopilot create --title "Daily triage" --agent <slug> --mode create_issue --description "..."
multica autopilot trigger-add <id> ...   # attach a schedule
multica autopilot trigger <id>           # run once now
multica autopilot runs <id>

Only --mode create_issue is fully supported end-to-end today; run_only is not.

Daemon (local runtime)

multica daemon start            # background; add --foreground to attach
multica daemon status
multica daemon logs
multica daemon stop

If an assigned issue isn't being picked up, multica daemon status first — local-runtime agents won't run with the daemon offline.

Global flags and env vars

Flag	Env var	Purpose
--workspace-id <id>	MULTICA_WORKSPACE_ID	Override active workspace
--server-url <url>	MULTICA_SERVER_URL	Point at a non-default server (self-hosted)
--profile <name>	—	Isolated config/daemon/workspace set (e.g. dev)
--output json	—	Machine-readable output for get/create/list

Discover more

Every command supports --help:

multica issue create --help
multica agent update --help

See REFERENCE.md for the full command tree, less-common subcommands (workspace admin, repo, attachment, config, runtime activity), and authentication notes.