ForgeOps connects your local AI models into autonomous pipelines — define tasks, set schedules, and let agents run, recover, and report without you in the loop.
# Connect local model, define tasks, go agent translator { model: "qwen3.6-35b@localhost:11434" schedule: "0 6 * * *" retry: 3 on_failure: "alert_slack" } pipeline daily_digest { translator → summarizer → notifier }
Cron-based or event-triggered pipelines. Agents wake up, run tasks, and report back — without you watching.
Ollama, Ollama API, LM Studio, or any OpenAI-compatible endpoint. Your models, your infra.
Agents detect failures, retry with backoff, and alert you only when truly stuck. Silent failures are gone.
Persistent agent memory across runs. No context loss between sessions. Agents pick up where they left off.
String multiple agents into pipelines. Output from one agent feeds into the next. Fully composable.
Everything runs on your infrastructure. No data leaves your network. For developers who care about privacy.
Write a simple config — model endpoint, task description, tools, and schedule. No boilerplate.
Point to Ollama, LM Studio, or any local LLM. ForgeOps handles the connection and keeps it alive.
Schedules trigger agents. Pipelines chain. Failures retry. You get the output, not the work.
The gap between "running an AI agent" and "having an AI employee" is orchestration.
ForgeOps closes it.