agents-cli
/agents-cli
Run and manage many AI coding agents (Claude Code, Codex, Cursor, Gemini/Antigravity, Grok, Kimi, OpenCode, Droid, …) from one CLI — the `agents` command. Use this to run multiple coding agents in parallel, manage multiple Claude Code accounts, keep working when I hit my usage limit, resume a session on another machine, or pin the agent CLI version. Triggers on: run multiple coding agents in parallel, manage multiple Claude Code accounts, I hit my usage limit, resume a session on another machine, pin the agent CLI version.
- Author
- phnx-labs ↗
- Repository
- agi-cli ↗
- Licence
- Apache-2.0
- Last repository change
- 27.10.2026
Assessment and security
- Files in this package
- 1
- Scripts
- 0
- HTML comments
- 0
- URLs in SKILL.md
- 1
No signals detected. Repository changed on 27.10.2026, without references to retired models or APIs.
Apache-2.0, otwarta: można używać i zmieniać, z zachowaniem licencji i informacji o zmianach.
What it does and when to use it
Run and manage many AI coding agents (Claude Code, Codex, Cursor, Gemini/Antigravity, Grok, Kimi, OpenCode, Droid, …) from one CLI — the `agents` command. Use this to run multiple coding agents in parallel, manage multiple Claude Code accounts, keep working when I hit my usage limit, resume a session on another machine, or pin the agent CLI version. Triggers on: run multiple coding agents in parallel, manage multiple Claude Code accounts, I hit my usage limit, resume a session on another machine, pin the agent CLI version.
Arguments
/agents-cli [teams|accounts|run|sessions|use]
Original author description
Run and manage many AI coding agents (Claude Code, Codex, Cursor, Gemini/Antigravity, Grok, Kimi, OpenCode, Droid, …) from one CLI — the `agents` command. Use this to run multiple coding agents in parallel, manage multiple Claude Code accounts, keep working when I hit my usage limit, resume a session on another machine, or pin the agent CLI version. Triggers on: run multiple coding agents in parallel, manage multiple Claude Code accounts, I hit my usage limit, resume a session on another machine, pin the agent CLI version.
How it works
- Run multiple coding agents in parallel
- Manage multiple Claude Code accounts
- I hit my usage limit
- Resume a session on another machine
- Pin the agent CLI version
- More
SKILL.md preview
--- name: agents-cli description: "Run and manage many AI coding agents (Claude Code, Codex, Cursor, Gemini/Antigravity, Grok, Kimi, OpenCode, Droid, …) from one CLI — the `agents` command. Use this to run multiple coding agents in parallel, manage multiple Claude Code accounts, keep working when I hit my usage limit, resume a session on another machine, or pin the agent CLI version. Triggers on: run multiple coding agents in parallel, manage multiple Claude Code accounts, I hit my usage limit, resume a session on another machine, pin the agent CLI version." argument-hint: "[teams|accounts|run|sessions|use]" allowed-tools: Bash(agents*) user-invocable: true --- # agents CLI `agents` (alias `ag`) is one control plane for every AI coding-agent harness — Claude Code, Codex, Cursor, Antigravity (Gemini), Grok, Kimi, OpenCode, Droid, OpenClaw, and more. Install it once and it manages versions, accounts, parallel runs, and cross-machine sessions for all of them. ```bash npm install -g @phnx-labs/agents-cli # provides `agents` and `ag` agents --version # 1.22.58 ``` Every recipe below is a real command. Run `agents <group> --help` for the full flag list — each group ships a workflow-first help block, not a flag dump. --- ## Run multiple coding agents in parallel `agents teams` runs several agents at once on one shared task, each isolated in its own git worktree, wired into a dependency DAG. ```bash # 1. Create a team agents teams create pricing-page # 2. Add teammates — different harnesses, run in parallel, each in its own worktree agents teams add pricing-page claude "Rewrite the /v2/pricing endpoint" --name backend agents teams add pricing-page codex "Build the /pricing route, three-tier layout" --name frontend # 3. DAG dependency — QA waits for BOTH to finish (--after) agents teams add pricing-page claude "Run the Playwright suite, fix flakes" --name qa --after backend,frontend # 4. Start everyone (respects --after) and watch live agents teams start pricing-page --watch # 5. Check in without re-reading everything agents teams status pricing-page # 6. Wind down when shipped agents teams disband pricing-page ``` For a single one-off agent (not a team) use `agents run <agent> "task" --mode edit`. To fan the same task across your *machines* rather than agents, add `--device auto` (see below). --- ## Manage multiple Claude Code accounts Each installed harness version can carry its own login, and named accounts let you pick which one a run uses. `agents accounts` is the surface. ```bash # See every native login and named account bundle agents accounts list # Add a second named login for a harness (runs native login into its own slot) agents accounts add claude work # Run on a specific account for just this one run agents run claude --account work "audit the auth middleware" # Set the default account for a harness agents accounts default claude work # Spread load automatically across the accounts you're signed into agents run claude -b "fix the failing test" # -b = --strategy balanced ``` `agents view` shows every installed version, its bound account, and that account's live usage bars side by side, so you can see which login has headroom before you run. --- ## I hit my usage limit Three built-in ways to keep working when a harness or account is rate-limited — no manual babysitting. ```bash # 1. Automatic harness fallback: if Claude rate-limits, retry on Codex, then Antigravity agents run claude --fallback codex,antigravity "finish the refactor" # 2. Balanced account rotation: pick the signed-in account you've used least recently agents run claude -b "keep going" # 3. Full-auto: pick the harness with the most account headroom AND a fresh account agents run auto "fix the flaky test" --mode edit ``` Check headroom first with `agents view` — the `S:` (5-hour) and `W:` (weekly) bars per account show exactly what's left before you switch. --- ## Resume a session on another machine Sessions sync across your fleet, so a conversation started on one box is searchable and resumable from any other. Resume is machine-aware: it hops to the device that owns the live harness state for you. ```bash # Find the session (the standalone sessions CLI; add --host <box> to search another machine) sessions "auth middleware" # Resume by id or short prefix — from ANY machine; it routes to the owning device agents run auto --resume 019fd0c8-b3e9-77a2-a1a4-444698c4d897 agents run auto --resume 019fd114 "now wire up the tests" ``` `agents run auto --device auto "task"` picks the least-loaded worker in your fleet and runs there; `agents devices` lists the machines you can reach. --- ## Pin the agent CLI version Pin a harness version globally or per-project so every run is reproducible. `add` installs; `use` sets the default (they're separate on purpose). ```bash # Install a specific version (reproducibility) agents add claude@2.1.112 # Make it the global default (only `use` sets a default) agents use claude@2.1.112 # Pin THIS project only — written to the project's .agents/agents.yaml agents use claude@2.1.112 --project # Install a clean, isolated copy that never touches your existing setup agents add claude@2.1.112 --isolated ``` `agents add claude@latest` / `@oldest` resolve symbolic versions. Each pinned version runs in its own isolated home, so configs and logins never bleed between releases. --- ## More | You want to… | Command | |---|---| | See installed agents + usage | `agents view` | | Register an MCP server for every harness at once | `agents mcp add` | | Drive a real browser (fill forms, screenshot) | `browser` (standalone CLI; `agents setup browser` installs it) | | Store credentials, inject into runs | `secrets` (standalone CLI) + `agents run --secrets <bundle>` | | Schedule recurring agent jobs | OS cron / systemd timer / launchd calling `agents run` | | Sync skills/commands/rules across the fleet | `agents sync` | Full docs: <https://github.com/phnx-labs/agi-cli>. Every group teaches its own workflow — start with `agents <group> --help`.
Versions
The database has no other skills named agents-cli.
Database snapshot: 13.11.2026 · package version b0d0e4a7dd49 (SHA-256 digest of the entire package)
Downloads in the last 30 days: 1 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 90handoff
Keep a handoff file so another coding agent (Claude Code, Codex, Kimi Code, Gemini CLI, Cursor) or a later session can continue this task without re-explaining. Use when the user asks to start, update or resume a handoff, says they will switch agents, when… - 80cmk:agent-vendors
Use when the user asks to "vendor the kit's skills into this repo", "set up agent adapters", "add a coding-agent vendor", "wire skills for Claude Code / Codex / OpenCode / Grok Build / Cursor", or whenever a repo needs one canonical skill set discoverable by multiple coding agents. - 100use-coding-agents
How to use the coding agents installed on this machine (claude, codex, droid, grok, hermes, opencode, copilot, pi) as plain sub-agents in orchestrated workflows: one headless invocation, one prompt in, one result out. Deliberately ignores each CLI's own orchestration features… - 90workflow
Create, run, monitor, cancel, or resume deterministic JavaScript workflows that orchestrate multiple Codex, Claude, Cursor, or Kimi agents. Use for parallel investigation or review, staged agent pipelines, resumable long-running work, or when the user explicitly asks for a… - 100parallel-agents
Coordinates several coding agents (Claude Code, Codex, Cursor, Gemini CLI, or several copies of one) working in the same repository at the same time, so they never edit the same files or run clashing commands. Use before editing files whenever other agents may be active in this… - 91send
Use when a user wants to start a new Claude Code, Codex, OpenCode, Cursor CLI, Pi, or Grok session from a coSlash session.