Learn how a concrete helpdesk actually handles tickets by deriving recurring ticket families, handling playbooks, state norms, note-style references, and fast retrieval artifacts from a historical ticket dataset.
Use this skill when CTOX should learn how a specific helpdesk works from a historical ticket export.
This skill is not about generic clustering. It is about deriving an operating model that CTOX can reuse while working future tickets in the same desk.
The goal is to produce:
Read these first:
The run is only acceptable if it produces:
operating_families.jsonfamily_playbooks.jsonstate_transition_norms.jsonnote_style_refs.jsonretrieval_index.jsonloperating_model.mdIf embeddings are enabled, also produce:
retrieval_vectors.npyEvery promoted family playbook must contain a decision_support block with:
operator_summarytriage_focushandling_stepsclose_whencaution_signalsnote_guidanceThe output must answer, for repeated ticket families:
Build the operating model:
python3 skills/system/ticket-operating-model-bootstrap/scripts/build_ticket_operating_model.py \
--input-xlsx <path> \
--output-dir <dir> \
--top-families 20 \
--min-family-size 20
Refine the strongest families into operator-facing decision support with gpt-5.4-nano:
OPENAI_API_KEY=... python3 skills/system/ticket-operating-model-bootstrap/scripts/build_ticket_operating_model.py \
--input-xlsx <path> \
--output-dir <dir> \
--top-families 20 \
--min-family-size 20 \
--openai-model gpt-5.4-nano \
--openai-refine-limit 6
Add retrieval vectors when the host can sustain local embeddings:
python3 skills/system/ticket-operating-model-bootstrap/scripts/build_ticket_operating_model.py \
--input-xlsx <path> \
--output-dir <dir> \
--top-families 20 \
--min-family-size 20 \
--embedding-provider sentence-transformers \
--embedding-model Qwen/Qwen3-Embedding-0.6B
Query the resulting operating model for a new ticket:
python3 skills/system/ticket-operating-model-bootstrap/scripts/query_ticket_operating_model.py \
--model-dir <dir> \
--query "<new ticket text>" \
--top-k 8