Problem 01
Refund skill
Check order state, policy, payment boundary, and approval requirement.
Support behavior
Reusable AI behavior for specific support situations.
Skills shape how Aura handles refunds, bugs, angry customers, missing information, fallback, and handoff while staying tied to sources and approvals.
TriggerRetrieve sourceDecide modeReview outcome
Support playbooks
repeatable support steps
SOP behavior
Support skill library
version controlled
Reusable skills
Fallback protocol
when confidence < 70%
enabled
Frustrated customer
de-escalation brief
reviewed
Bug escalation
engineering packet
template v1.7
Aura behavior
Tone
calm, direct, policy-aware
team SOP
Evidence
must cite approved source
strict
Fallback
handoff to human owner
required
Governance
Owner
Support ops
approves changes
Version
SOP-2026.04
active
Audit
skill applied to REP-1842
visible
library01
Select skill
Operators attach a behavior pattern to a queue or workflow.
runtime02
Apply SOP
Aura follows the approved fallback and escalation instructions.
audit03
Inspect
The applied skill version is stored with the conversation.
Tools overview
Skills
Skills shape how Aura handles refunds, bugs, angry customers, missing information, fallback, and handoff while staying tied to sources and approvals.
How it works
repeatable support steps
01
Trigger
Intent, sentiment, topic, or channel selects a skill.
02
Retrieve source
Skill pulls approved process docs or policy material.
03
Decide mode
Draft, ask a question, prepare action, or escalate.
04
Review outcome
Managers inspect performance, overrides, and gaps.
Operational capability
Support quality depends on consistent judgment in repeated cases.
Skills give Aura a repeatable policy-aware pattern for known support scenarios.
Problem 02
Bug escalation skill
Collect reproduction steps, media, environment, and impact.
Problem 03
Frustrated customer skill
Detect sentiment, lower automation, and route to a human quickly.
Skill lifecycle
Behavior tied to sources and ticket status.
Detect
01Trigger
Intent, sentiment, topic, or channel selects a skill.
Ground
02Retrieve source
Skill pulls approved process docs or policy material.
Control
03Decide mode
Draft, ask a question, prepare action, or escalate.
Improve
04Review outcome
Managers inspect performance, overrides, and gaps.
Skill model
Each skill needs a clear boundary.
Skill
Data used
Boundary
Skill
Refund
Data used
Order, payment, policy
Boundary
Approval before action
Skill
Bug
Data used
Transcript, media, environment
Boundary
Technical handoff
Skill
Frustration
Data used
Sentiment, history, priority
Boundary
Human route
Skill
Missing info
Data used
Required fields and source policy
Boundary
Ask before guessing
Related pages
See what to read next.
Tools
MCP Tools
Approved tools let Aura and operators query context, prepare updates, trigger workflows, and package actions while permissions and approvals decide what can run.
Tools
Integrations Hub
The Integrations Hub shows what Replofy can read, where conversations enter, which tools Aura can use, and how connection health is monitored.
Tools
SDKSoon
The SDK/API surface is for teams that need ticket creation, workflow triggers, analytics reads, and internal support operations inside product or admin tools.
Design skills
Standardize how Aura behaves in repeatable situations.
Encode process docs, fallback rules, and escalation paths without removing human control.