Skills CX operations
Building knowledge from resolved conversations
The best article briefs already exist — in tickets your agents closed correctly last week. Most teams instead brainstorm from a conference room, write content customers never asked for, and wonder why deflection flatlines.
npx rulebase-skills install cx-knowledge-from-conversationsWhen to use it
Reach for this when someone says any of these — they are the phrases the skill itself triggers on:
- “what articles should we write from tickets”
How it works
The method, in the order the skill runs it. The full procedure — tables, worked examples and the edge cases — is in the skill itself.
Start from repeat drivers, not random samples
Do not mine "interesting tickets." Mine high-volume, high-repeat contact drivers where resolved conversations show a stable answer: Filter to drivers with consistent resolution.
What to extract from each conversation
For each sample, record: Customer question, Resolution steps, Boundary conditions, Links and artefacts, Macro or snippet used.
Draft structure before prose
Do not paste agent replies into the help centre. Convert to customer-facing structure.
PII and compliance gate (mandatory before publish)
Every draft from conversations must pass.
Review before publish
Minimum reviewers: Topic owner (policy/product), Support lead or senior agent, Optional.
Ongoing mining rhythm
Agents who repeatedly write the same bespoke reply are writing your next article. Promote those patterns deliberately.
Related skills
Free and open source, and vendor-neutral — it reads the conversations from whichever helpdesk you already run. Browse all 149 skills · connect your helpdesk over MCP · source on GitHub
