Skip to content →
Core workflowsFailure has consequences
GitHub
Core workflows

Failure has consequences

See how outages, invalid requests, and unanswered queues affect people in the world.

The human cost

Praxis translates technical failures into changes to the affected actor. Stress rises, trust falls, and the relationship can degrade from active to at-risk to lost.

SeverityEffect
MinorStress +0.10, trust −0.04, active relationships become at-risk
MajorStress +0.18, trust −0.08, an already at-risk relationship can become lost

Connector failures

HTTP and configuration errors are classified before consequences are applied.

SignalClassificationRetryable
HTTP 5xx or network TypeErrorsystem-unavailableYes
HTTP 401 or 403access-deniedNo
HTTP 400, 404, 409, or 422invalid-requestNo
Missing connector, token, or capabilityconfiguration-neededNo
Anything elseunknownNo

Unanswered work

Pending organic requests also age against intent-specific deadlines. Triage starts timing out after 30 minutes, clinical updates after 90, bookings and emotional messages after 120, and other work after 180.

At the first deadline Praxis marks the actor at-risk and creates rescue work. At twice the deadline it fails the decision, sets stress to 1, reduces trust, marks the actor lost, and creates urgent recovery work.

Testing failure paths

Select mock-random for the capability you want to exercise, then create targeted work through the panel or API. The request, operation attempt, error classification, task, event, and profile consequence will be visible in one world state.

Configure message failure testing
curl -X PATCH http://localhost:3000/api/simulation \
  -H 'Content-Type: application/json' \
  -d '{"config":{"transports":{"message":"mock-random"}}}'