incident-response-loop
incident-response-loop is a publishable AgentPM Loop package for assess, execute, review, and revision-driven incident follow-through.
It packages the portable orchestration contract for a common incident workflow:
- first assess the current incident state and ownership
- then move into active execution when enough context exists
- then review whether the incident can close, needs another execution pass, or should hand off
The goal is to show that a Loop can capture iterative incident control flow without embedding runtime code, provider behavior, or package-specific execution logic.
Use it for:
- demonstrating a clear assess -> execute -> review cycle with a reachable return path into execution
- showing a realistic operations Loop that uses bounded steps, access intent, and fail-phase error handling
- seeding a public incident-response example that fits naturally with the existing ops agent, skill, profile, and memory story
The package includes:
- agent.json
- README.md
Local development
From this directory:
agentpm lint
agentpm publish --dry-run
The source for this Loop package can be found here.