Skip to main content
Let an AI agent automatically diagnose and propose fixes when this schedule fails. When a run fails on a schedule with Self-Healing enabled, Paradime triggers a DinoAI agent against the failure. The agent diagnoses the run, implements a fix, and opens a pull request for your team to review. The healing session is always readable in Agent Studio, tagged with the Bolt source. If the schedule also has a Slack channel configured, the agent works inside that failure thread as well.
You stay in control. The agent always opens a PR, it never merges to your default branch. Reviewers approve the fix before it reaches production.
Prerequisites
  • A programmable agent defined under .dinoai/agents/<name>.yml on the schedule’s Git branch. This is the agent that diagnoses and fixes the run, and it is the only requirement.
  • A Slack or Microsoft Teams channel is optional. Configure one under Notifications if you want the healing session to appear in chat as well as in the app.

​
Enable from the Bolt UI

In the schedule’s Self-healing section, selecting an agent enables self-healing:
  1. Select an Agent name, the DinoAI programmable agent that diagnoses and fixes the run. The list is populated from .dinoai/agents/*.yml on the schedule’s Git branch.
  2. Optionally select a Channel. If you pick a Slack channel, it must already be configured as a Slack notification channel on this schedule.
  3. Save the schedule.
To turn self-healing off, clear the selected agent.

​
Configure from YAML

If you manage schedules as code, add a self_healing block to the schedule. Only enabled and agent_name are required:
Add slack_channel when you want the agent to thread into a chat notification:
If you do set self_healing.slack_channel, it must also appear in notifications.slack_channels for the same schedule. The validator rejects the schedule otherwise, because the agent threads into the failure notification and that notification has to exist for the agent to find its thread. Leaving slack_channel out entirely is valid, and the agent then runs without a thread.

​
What happens when a run fails

Without a channel configured
  1. Paradime triggers the DinoAI agent against the failed run.
  2. The session appears in Agent Studio tagged with the Bolt source, with the full transcript, status, and title.
  3. The agent ends with a PR link for the proposed fix.
With a Slack channel configured
  1. The standard failure notification posts to every Slack channel in notifications.slack_channels.
  2. In the Self-Healing channel, Paradime posts 🦖 Self-healing enabled — starting healing session... into the failure thread and triggers the DinoAI agent.
  3. In every other notification channel, Paradime posts 🦖 Self-healing started in #<healing-channel> so reviewers know where to follow along.
  4. The agent works in-thread, posts progress, and ends with a PR link for the proposed fix.
With a Microsoft Teams destination Paradime posts a “self-healing started” card to the Teams webhook and runs the agent, with the session readable in Agent Studio. In every case, a failed run produces one healing session, even when the failure alerts several times in the same channel. If the same schedule has been self-healed recently, the agent is given a summary of those prior sessions so it doesn’t duplicate work, for example if an open PR already proposes a fix. For the full agent-side narrative, see DinoAI → Self-Healing.

Define this as code

Configure the self_healing block (enabled, agent_name, and an optional slack_channel).