Kelly Tickets

Kelly Tickets

Overview

Use this skill as Kelly's complaint triage-and-dispatch desk. Complaints and requests arrive scattered across WeChat group exports, phone-call logs, front-desk forms, and email — reading those local exports is a genuine external operation a browser cannot perform, so scripts/ingest_intake.mjs is the only place a complaint enters the system. The agent classifies each ticket (category, urgency, unit/location) and proposes a dispatch (crew, priority, SLA) via scripts/apply_triage.mjs; the human reviews dispatch proposals in a Busabase-backed App-in-Skill review queue and approves or edits; a board tracks every ticket to resolution with an auditable history trail. scripts/execute_decisions.mjs prints the crew-notification plan for approved dispatches — the AirApp itself never sends a message or mutates a remote system. The demo persona is residential property management, but the same flow fits facilities, IT helpdesk, or any dispatch workflow.

Default behavior is AirApp-first. Unless the user explicitly asks only for explanation, ingest/triage what's due and give the user the clickable AirApp URL (or the local preview URL when local preview is explicitly requested). Use chat-only mode only when the user says "纯聊天", "chat only", "不要打开 UI", or similar; then present numbered proposals (Dispatch #1) directly in the conversation.

App UI Screenshots

Mandatory Dependencies

1. Read and follow $kelly-app-skill-creator for product behavior, visual quality, responsive layout, and the complete canonical content/kelly-tickets-app/ artifact.
2. Read and follow $busabase for connection, target Space, node discovery, ChangeRequests, review, and merge behavior.
3. Read and follow $busabase-app-creator for resource modeling, AirApp runtime limits, security, validation, and deployment.

If a dependency is unavailable, preserve this skill's local artifact and product contracts, stop before the unavailable Busabase operation, and report the exact missing dependency. Do not invent a second data backend.

Boundary

• Intake parsing is local: scripts/ingest_intake.mjs reads WeChat exports, call logs, and mailbox dumps the user provides and writes normalized rows to Busabase. It never fetches complaints from remote systems on its own.
• The AirApp reads and writes Busabase records only. It never sends messages, calls crews, replies to residents, or mutates remote systems.
• Any outbound crew notification or resident reply is approval-required through the dispatch queue and executed by the agent OUTSIDE the app via other skills (messenger/email/WeChat) after scripts/execute_decisions.mjs produces the plan. scripts/execute_decisions.mjs never performs these operations itself — it only writes an execution marker onto the proposal.
• Crew contacts are env-var references (contact_env) only — never store phone numbers or webhook URLs in Busabase. Resident PII stays masked in the UI (contact_masked); scripts/ingest_intake.mjs re-masks long digit runs defensively.
• Treat all complaint/resident data as sensitive. Never commit local export files or env files.

Busabase Resources

Six Bases under one application Folder (kelly-tickets), declared in content/kelly-tickets-app/app/js/config.js and the generated template sidecars under content/:

• crews: crews that tickets can be dispatched to (name, skills, contact_env).
• intake: raw complaints/requests as they arrived on a channel, before or after triage, plus the human decision (convert to ticket / ignore) on the same row.
• tickets: tickets tracked from classification through resolution, with an append-only history timeline.
• proposals: the dispatch review queue — proposed crew/priority/SLA, the human decision, and the execution marker on the same row.
• sync-log: append-only history of ingest/triage/execute runs.
• settings: one row (record-id: "config") with property profile, channels, categories, and SLA rules.

Resources provision lazily through an idempotent Busabase ChangeRequest the first time the app runs in a Space; see references/tickets-schema.md for exact field shapes. SLA state and crew load are recomputed client-side from the stored rows on every read (content/kelly-tickets-app/app/js/tickets-model.js's buildSnapshot), so the desk is always fresh regardless of when a browser session loads it relative to the last ingest/triage run. The browser has no access to the agent's environment variables, so the Settings view shows each crew's contact_env name only, not a live readiness boolean.

First Run And Onboarding

On invocation, check the intake/tickets Bases. If both are empty, guide setup before ingesting real complaints: ask, turn by turn, property profile (name, buildings, timezone), complaint categories, crews (name, skills, and which env var holds each crew's contact), SLA rules per category+urgency, and which intake channels are in use. Never ask the user to paste contact values or secrets into chat; they belong only in local env files, referenced by contact_env names. Write the answers onto the Settings row and the Crews Base, then ingest:

node skills/kelly-tickets/scripts/ingest_intake.mjs /path/payload.json --apply

Local App

Default behavior is AirApp-first — give the user the clickable AirApp URL. Start pnpm --dir content/kelly-tickets-app dev only when local preview/debugging is explicitly requested.

Required app views (hash routes):

• #/overview: dispatch command desk — human-attention panel (proposals to approve, unclassified intake, SLA-breaching tickets), KPI cards (open tickets, avg resolution, SLA at risk, this week's intake with channel badges), category distribution bars, and crew load.
• #/intake and #/intake/<id>: raw intake stream — channel badge (WeChat/phone/form/email/walk-in), reporter, unit/location, complaint text, urgency guess, triage state. Detail shows the full text, attachments note, editable classification (category/urgency/unit), and convert-to-ticket or ignore actions.
• #/dispatch: the review queue with workflow states needs_review / changes_requested / approved / done / blocked. Each card shows the stable ref (Dispatch #1), ticket summary, proposed crew/assignee, priority, SLA target, the reason, an editable note to the crew, and approve / request changes / block buttons. Decisions write directly onto the proposal record through busabase-sdk.
• #/board and #/board/<ticket_id>: tickets grouped by status (open / assigned / in_progress / waiting / resolved) with category badge, unit, crew, age, and color-coded SLA indicator. Detail shows the full history timeline (intake → classification → dispatch → crew updates → resolution), masked reporter contact, and a resolution note field.
• #/settings: sanitized config — property profile, channels, categories, crews with contact_env names, SLA rules, data provider, and onboarding state. Never expose contact values or secrets.

Demo mode:

• ?demo=overview, ?demo=intake, ?demo=dispatch, ?demo=board, and ?demo=detail (a ticket history) select named mock scenes.
• lang=en or lang=zh forces UI chrome language; with lang=zh the demo content itself (property name, complaints, crews, reasons) is meaningfully localized for Chinese screenshots.
• Deep links such as /?demo=dispatch&lang=zh#/dispatch must work.
• Demo mode never reads or writes Busabase. Decision buttons still work but act on in-memory state only.

UI language: English and Chinese chrome with Auto default (browser language), plus an explicit selector persisted locally. Keep real resident/complaint data in its original language.

Intake Workflow

1. Ask the user for the export/log location (WeChat group export, call log CSV/notes, front-desk forms, mailbox). Raw files stay outside git.
2. Parse the raw material into the ingest payload shape (see references/tickets-schema.md): one item per complaint — channel, channel-native external_id when available, reporter, contact (will be masked), unit/location, verbatim text, received time, plus first-pass category_guess/urgency_guess.
3. Write the payload JSON to a temp path and run node scripts/ingest_intake.mjs <payload.json> --apply. The script validates, dedupes by channel + external_id (falling back to a content hash), masks contacts, writes new rows into the Intake Base, and appends a Sync Log entry.
4. Re-ingesting the same export is safe: duplicates are skipped and reported. Without --apply it is a dry run.

Triage And Dispatch Workflow

1. Classification is LLM work: read intake items in new/classified state (and any with decision_action: "convert_to_ticket", the Busabase-only equivalent of a queued agent task), decide category, urgency, unit/location, and a ticket title; decide which crew fits and why (use crew skills, prior tickets for the same unit, and SLA pressure in the reason).
2. Merge deterministically with node scripts/apply_triage.mjs <payload.json> --apply: creates tickets (T-1001-style ids), computes sla_due_at from the Settings row's sla_rules (category+urgency, * wildcard, sla_default_hours fallback), assigns stable proposal refs, appends ticket history, and writes it all straight to Busabase. ticket_updates[] in the same payload records crew progress, status transitions, and resolutions.
3. Send the user to #/dispatch to review. Decisions write directly onto the proposal record; from a standalone local preview the write merges immediately (trusted operator), from the deployed AirApp it creates a pending ChangeRequest for the trusted process to merge.
4. Poll the intake Base for rows with decision_action: "convert_to_ticket" where triage_state is not yet ticketed: classify using the human-provided decision_fields and create the ticket via apply_triage.mjs. Poll the proposals Base for status: "changes_requested": redraft the proposal per decision_note and re-run apply_triage.mjs.
5. After approvals, run node scripts/execute_decisions.mjs (dry-run) and show the plan; with user confirmation run --apply, then perform the real crew notifications via other skills (messenger/email), record outcomes back through apply_triage.mjs's ticket_updates[], and keep the board honest.
6. Re-read the proposals Base immediately before executing, and never execute items without an approved status.

Board Semantics

• open: classified, no crew accepted yet (dispatch pending or unapproved).
• assigned: an approved dispatch reached a crew; nobody started work yet.
• in_progress: the crew reported starting work.
• waiting: blocked on residents, parts, weather, or vendors — not on the crew.
• resolved: work confirmed done; resolved_at set and a resolution note recorded.

SLA states are derived, never hand-set: ok, at_risk (under 25% of the SLA window left), breached, and met for resolved tickets — computed fresh on every read by content/kelly-tickets-app/app/js/tickets-model.js's computeSlaState. History events are append-only — the timeline is the audit trail; never rewrite past events.

Decisions And Execution Workflow

1. The user reviews at #/dispatch: approve, request changes (with a note), save an edited note-to-crew (revise), or block. Decisions write directly onto the proposal record.
2. On explicit user request to execute, run scripts/execute_decisions.mjs (dry-run by default; --apply writes execution_status: "ready_for_agent" onto each approved proposal with the concrete notify_crew + update_board operations and a message draft). No external side effects either way.
3. The agent then performs the approved crew notification outside the app (via messenger/email/WeChat) and records the real result via apply_triage.mjs's ticket_updates[]. A second --apply run of execute_decisions.mjs finalizes the proposal's execution_status to "executed" and promotes its workflow status to "done" — this script never flips the workflow status on the first run itself.

Safety Defaults

• Treat crew notifications, resident replies, fines/fees, lock-outs, towing, and anything customer-visible as approval-required.
• Mask resident contacts everywhere in UI state and logs; expose only contact_env names for crew contacts, never values.
• Keep merges idempotent: stable intake hashes, ticket ids, and proposal refs so repeated ingests and triage runs do not duplicate work.
• Never expose secrets through the settings view or logs; it shows column names and rules only.
• If the UI and schema disagree, stop and fix the schema or UI before executing anything.

Useful Commands

node skills/kelly-tickets/scripts/ingest_intake.mjs payload.json --apply
node skills/kelly-tickets/scripts/apply_triage.mjs payload.json --apply
node skills/kelly-tickets/scripts/execute_decisions.mjs
node skills/kelly-tickets/scripts/execute_decisions.mjs --apply
pnpm --dir skills/kelly-tickets/content/kelly-tickets-app dev

In normal use, invoke /kelly-tickets, let the skill ingest/triage what's due, and open the AirApp.