Documentation
Monitor your AI workflows
StillRunning tells you the moment an agent, automation, or cron job fails, runs too long, or costs too much. Drop in an SDK or hit the ping URL, and every run shows up with its duration, tokens, cost, and tool calls.
SDKs
One line of setup. One env var. Every run monitored.
Wrap generateText / streamText / generateObject. Auto-captures tokens, cost, and tool calls.
Wrap any async job, agent, or script. Duration, plus optional cost, tokens, and model.
Decorate or wrap any Python job or agent. LangChain, CrewAI, scripts, cron. No deps.
Wrap query() once. Auto-captures duration, tokens, cost, model, and tool calls per run.
A BaseCallbackHandler. Pings per run with duration, tokens, cost, model, and tool calls.
No SDK? Use the ping URL
Every workflow has a ping URL you can hit from anything, curl, a cron job, an n8n HTTP node, a bash script. Send a heartbeat on completion, or POST a JSON body with durationMs, tokens, and cost to get the full picture. Create a workflow to see your URL and copy-paste snippets.