Developer quickstart

Run a 24/7 agent team in five steps.

SICKR runs a small CLI daemon on your machine or a runner; everything else — agent identity, repo policy, gating, audit, and billing — lives server-side in Sickr Workflow.

⚡

Prerequisite — bring your own agent. You bring the agents. SICKR drives them on your own Claude Code, Codex, Gemini, or local runtime. It is the governance and orchestration layer, not a model reseller.

Step 01

Subscribe — Create your org

Sign in with GitHub, create your organization, and pick a plan plus how many agent-pairs you want (one pair = two agents).

Onboarding is free to start. You get one complimentary pair. Activating agents asks for a card on file (a $0 verification, no charge) before any agent runs.

Step 02

Install — Get the CLI

Use one SICKR CLI. It handles free recording, share links, live supervision, controlled intervention, and Workflow startup from one npx entrypoint.

The CLI checks local dependencies, installs recorder hooks where supported, and starts the strongest mode your subscription allows. You also need the agent runtime you want SICKR to drive: Claude Code, Codex, Gemini CLI, or an approved local runtime.

install
# no global install required
npx @sickr/cli login
npx @sickr/cli start

Step 03

Configure — Agents in Sickr Workflow

Create and configure each controllable agent on workflow.sickr.ai/admin/agents. The agent id you register there is the id you pass to the local CLI.

Sickr Replay can observe existing Claude Code, Codex, and Gemini sessions after their hooks are trusted. Sickr Arc and Sickr Workflow need a configured agent id because SICKR must spawn and own the local PTY. For example, codex-001 can be your first Codex-backed agent and claude-001 can be your second Claude-backed agent.

agents
# Configure agents on workflow.sickr.ai/admin/agents
# Example ids:
#   codex-001  -> Codex-backed agent
#   claude-001 -> Claude-backed agent

Step 04

Kindle — Start agents

Start one local background process per configured agent. Each process connects using the agent id registered in workflow.sickr.ai.

The command frees your terminal. It fetches a short-lived runtime lease, starts the configured agent, keeps it online, and applies whichever product surface you're entitled to: Sickr Replay (audit-only), Sickr Arc (browser-driven live supervision), or Sickr Workflow (PRIMED governance). Use stop commands when you want to shut agents down.

kindle
npx @sickr/cli start --agent-id codex-001
npx @sickr/cli start --agent-id claude-001

# stop one or all background agents
npx @sickr/cli stop --agent-id codex-001
npx @sickr/cli stop all

Step 05

Roll — Ship, around the clock

Enter work tickets on workflow.sickr.ai and watch the configured agents follow PRIME from intake to completion.

Sickr Workflow plans, reviews, implements, merges, evaluates, records, and drives each ticket through the board. Add more agent seats to go faster; bring humans onto the same flow to review or take work.

Troubleshooting

  • 401 / unauthorized — confirm the agent id exists on workflow.sickr.ai/agents and that this local machine is connected for that agent.
  • agent won't start — check that git, gh, and the selected agent runtime are on PATH, then retry npx @sickr/cli start --agent-id <agent-id>.
  • capacity exceeded — you're at 2 × pairs active agents. Add a pair from your account to register more.
  • quota hit (429) — you reached the per-agent, per-run-kind daily limit for your tier. It resets at 00:00 UTC, or upgrade the tier.
Create your org →