Build a production incident-response agent Combine everything into an agent that investigates a large incident, uses external tools, waits safely for live updates, and proves its recommendations improved. AxFlow owns the fixed phases, AxAgent handles the investigation, and the runtime keeps long context and resumptions under control. python academy academy/capstone website/content-src/academy/course.mjs academy Build a production incident-response agent
ax
Python axllm
TypeScript Python Java C++ Go Rust
Final build · local execution

Build a production incident-response agent

Combine everything into an agent that investigates a large incident, uses external tools, waits safely for live updates, and proves its recommendations improved. AxFlow owns the fixed phases, AxAgent handles the investigation, and the runtime keeps long context and resumptions under control.

Build sequence
  1. Create the typed incident and resolution contracts.
  2. Use a Flow for intake, investigation, approval, and response phases.
  3. Keep the incident log runtime-only and add a persisted context map.
  4. Attach recorded MCP tools and an identity-aware event source.
  5. Observe progress, wake on an authorized resource update, and resume only the owned task continuation.
  6. Evaluate the baseline, apply an optimization artifact, and compare held-out results.
Recorded MCP fixture

Run the supporting demo locally

npm run example -- python src/examples/python/mcp/task-resume-flow.py

The browser never receives a provider key. Use the repository environment for provider-backed work and recorded events for deterministic protocol checks.

Architecture check

Connect the whole system