Support Console Operator
Resolve telecom support worklists by querying the support-console REST API and
producing structured JSON decisions. This skill covers ticket classification,
contact-center case handling, enterprise incident investigation, queue-quality
review, and mobile-data recovery.
Core Workflow
Every task follows the same three-phase pattern:
Parse input -- Read the payload file(s) provided in the prompt. Payloads
are CSV (ticket batches, queue snapshots) or JSON (case queues, worklists,
response requirements, customer preferences). Also read the answer template
JSON to know the exact output schema.
Query the API -- For every record in the payload, fetch relevant records
from the support-console API. Cross-reference multiple endpoints to build a
complete picture: tickets + diagnostics + outages, cases + customers + lines
- bills + plans + devices, incidents + export-runs + messages + SLA + accounts.
See api_endpoints.md for every endpoint.
Apply decision rules and emit JSON -- Classify each record using the
domain rules in decision_rules.md. Fill the
answer template exactly, preserving payload record order. Compute summary
counts from the decisions. Output only valid JSON conforming to the template;
do not wrap in markdown fences or add commentary.
Environment
The prompt provides a placeholder <TASK_ENV_BASE_URL>. Replace it with the
actual base URL from the environment (typically http://task-env:9003). No
credentials are needed. All endpoints are GET-only.
Task Types
The skill handles five task types, each with its own answer template and
decision surface. Read the payload carefully to determine which type applies,
then follow the matching rules in the decision-rules reference.
| Task |
Typical payload |
Key endpoints to query |
| Ticket batch |
ticket_batch.csv |
tickets, diagnostics, troubleshooting, outages, customers, accounts, lines |
| Contact-center queue |
case_queue.json |
cases, customers, lines, bills, plans, devices |
| Enterprise investigation |
client_complaint_email.txt + response_requirements.json |
incidents, export-runs, messages, accounts, SLA |
| Queue-quality review |
queue_snapshot.csv |
tickets, diagnostics, outages, accounts, customers, lines |
| Mobile-data recovery |
mobile_data_worklist.json |
lines, customers, bills, plans, devices |
Answer Templates
Every answer template is a JSON schema provided in the task payloads. Read it
before starting work. Templates define:
- Per-record decision fields (resolution status, actions, routes, blockers)
- Enumerated values for each field (see decision rules for when to use each)
- Summary aggregation fields (counts, totals)
Always preserve the order of records from the input payload in the output.
API Query Strategy
- Query one record at a time per endpoint. Use the id-specific endpoints
(
/api/tickets/{ticket_id}, /api/customers/{customer_id}, etc.) for
individual lookups after scanning list endpoints for matching ids.
- For tickets: start with
/api/tickets/{ticket_id}, then cross-reference
diagnostics, outages, and the account.
- For cases: start with
/api/customers/{customer_id} (derive from case_id
numbering), then fetch lines, bills, plans, and devices for that customer.
- For incidents: query
/api/enterprise/incidents/{incident_id}, then
export-runs and messages filtered by account and date range.
- For outages: query
/api/outages and match by service area or affected
accounts against the ticket/case record.
Naming Conventions
When output fields require generated names (channels, folders, report titles):
- Channel names: lowercase-hyphenated client name (e.g.
asteri-retail-inc)
- Evidence folders:
{Client Name} {Month Year} Investigation
- Report titles:
{Client Name} Export Failure - Resolution Report
Follow any explicit naming rules in the payload response_requirements.json.
Summary Computations
Compute summary fields by counting the decisions made:
- Count each status/route/action category from the per-record decisions
- For monetary totals, sum the
charge_amount_usd fields across all records
- For SLA credit percent, read the value from the SLA endpoint response
References
- api_endpoints.md -- Complete API endpoint catalog with response shapes
- decision_rules.md -- Domain-specific resolution logic for each task type
1---2name: support-console-operator3description: "Resolve telecom support tickets, contact-center cases, enterprise incidents, and mobile-data issues using a shared REST support-console API. Use when the task involves: (1) a batch of offline service tickets needing classification and routing, (2) a contact-center queue of mobile cases requiring self-service or billing actions, (3) an enterprise client export-complaint investigation with SLA and evidence handling, (4) a queue-quality review of tickets before SLA handoff, or (5) a mobile-data recovery worklist with refuel or carrier-update operations. Trigger on prompts mentioning support tickets, case queues, incident investigations, SLA reviews, or mobile-data recovery with a shared API endpoint."4license: MIT5compatibility: designed for deepagents-code6metadata:7 source: https://skillmd.com/skills/prism-shadow/support-console-operator8---910# Support Console Operator1112Resolve telecom support worklists by querying the support-console REST API and13producing structured JSON decisions. This skill covers ticket classification,14contact-center case handling, enterprise incident investigation, queue-quality15review, and mobile-data recovery.1617## Core Workflow1819Every task follows the same three-phase pattern:20211. **Parse input** -- Read the payload file(s) provided in the prompt. Payloads22 are CSV (ticket batches, queue snapshots) or JSON (case queues, worklists,23 response requirements, customer preferences). Also read the answer template24 JSON to know the exact output schema.25262. **Query the API** -- For every record in the payload, fetch relevant records27 from the support-console API. Cross-reference multiple endpoints to build a28 complete picture: tickets + diagnostics + outages, cases + customers + lines29 + bills + plans + devices, incidents + export-runs + messages + SLA + accounts.30 See [api_endpoints.md](references/api_endpoints.md) for every endpoint.31323. **Apply decision rules and emit JSON** -- Classify each record using the33 domain rules in [decision_rules.md](references/decision_rules.md). Fill the34 answer template exactly, preserving payload record order. Compute summary35 counts from the decisions. Output only valid JSON conforming to the template;36 do not wrap in markdown fences or add commentary.3738## Environment3940The prompt provides a placeholder `<TASK_ENV_BASE_URL>`. Replace it with the41actual base URL from the environment (typically `http://task-env:9003`). No42credentials are needed. All endpoints are GET-only.4344## Task Types4546The skill handles five task types, each with its own answer template and47decision surface. Read the payload carefully to determine which type applies,48then follow the matching rules in the decision-rules reference.4950| Task | Typical payload | Key endpoints to query |51|------|----------------|----------------------|52| Ticket batch | `ticket_batch.csv` | tickets, diagnostics, troubleshooting, outages, customers, accounts, lines |53| Contact-center queue | `case_queue.json` | cases, customers, lines, bills, plans, devices |54| Enterprise investigation | `client_complaint_email.txt` + `response_requirements.json` | incidents, export-runs, messages, accounts, SLA |55| Queue-quality review | `queue_snapshot.csv` | tickets, diagnostics, outages, accounts, customers, lines |56| Mobile-data recovery | `mobile_data_worklist.json` | lines, customers, bills, plans, devices |5758## Answer Templates5960Every answer template is a JSON schema provided in the task payloads. Read it61before starting work. Templates define:6263- Per-record decision fields (resolution status, actions, routes, blockers)64- Enumerated values for each field (see decision rules for when to use each)65- Summary aggregation fields (counts, totals)6667Always preserve the order of records from the input payload in the output.6869## API Query Strategy7071- Query one record at a time per endpoint. Use the id-specific endpoints72 (`/api/tickets/{ticket_id}`, `/api/customers/{customer_id}`, etc.) for73 individual lookups after scanning list endpoints for matching ids.74- For tickets: start with `/api/tickets/{ticket_id}`, then cross-reference75 diagnostics, outages, and the account.76- For cases: start with `/api/customers/{customer_id}` (derive from case_id77 numbering), then fetch lines, bills, plans, and devices for that customer.78- For incidents: query `/api/enterprise/incidents/{incident_id}`, then79 export-runs and messages filtered by account and date range.80- For outages: query `/api/outages` and match by service area or affected81 accounts against the ticket/case record.8283## Naming Conventions8485When output fields require generated names (channels, folders, report titles):8687- Channel names: lowercase-hyphenated client name (e.g. `asteri-retail-inc`)88- Evidence folders: `{Client Name} {Month Year} Investigation`89- Report titles: `{Client Name} Export Failure - Resolution Report`9091Follow any explicit naming rules in the payload `response_requirements.json`.9293## Summary Computations9495Compute summary fields by counting the decisions made:9697- Count each status/route/action category from the per-record decisions98- For monetary totals, sum the `charge_amount_usd` fields across all records99- For SLA credit percent, read the value from the SLA endpoint response100101## References102103- [api_endpoints.md](references/api_endpoints.md) -- Complete API endpoint catalog with response shapes104- [decision_rules.md](references/decision_rules.md) -- Domain-specific resolution logic for each task type