Decision controller ready.
Queue summaryActionable counts, evidence health, and execution phasesSecondary
Exact terminal source status, current revision, validation, and attestation must agree.
Action Board
Drag cards to change local planning lanes. The imported source bucket and status remain preserved for comparison and audit.
Sync Center
Import current statuses from CSV or JSON, link a local file for repeat refreshes, paste raw exports, or stage a read-only REST connector. Every import is previewed, hashed, diffed, validated, and reversible.
Upload a snapshot
Accepts CSV, TSV, JSON, and text exports. Automatic column detection is followed by a mapping and diff review.
Link a local source file
Use the browser File System Access API to refresh the same export later without choosing it again.
Paste an export
Paste CSV, TSV, or JSON from a ticket export. The parser detects the format and starts the same guarded import workflow.
Read-only connector studio
Prepare a generic REST or Zendesk JSON pull. Secrets remain session-only and are not included in exports.
| Result | Ticket | Customer | Changed fields | Incoming updated |
|---|
Use Link local file to retain a refreshable source handle when the browser supports it.
Ticket Intelligence
Classification, identifier extraction, data quality, risk reasons, closure gates, and escalation readiness are visible and inspectable. Derived values are labeled as derived and never replace source facts.
Queue Explorer
Use plain text or field queries such as bucket:pending, owner:name, module:dns, workflow:doing, changed:true, priority:1, or -status:solved.
| P | Ticket | Customer | Module | Source status | Effective lane | Owner | SLA | Risk | Workflow | Updated | Quality |
|---|
Build a verifiable evidence chain without uploading customer files.
Files are hashed locally. Only metadata and the fingerprint are retained. File contents are not stored or sent anywhere.
SHA-256 and metadata are calculated in this browser.
See the queue as a connected operational system.
Tickets are linked to customers, modules, owners, patterns, dependencies, and technical identifiers. The graph is derived from current source data and local planning state.
Model time, aging, capacity, and the next risk horizon.
The temporal model uses supplied timestamps when present and labels low-confidence estimates explicitly. It never fabricates customer timestamps or SLA deadlines.
| Ticket | Customer | Anchor | Age | Due horizon | Confidence | Risk | Effort | Next action |
|---|
Detect repetition before it becomes noise.
Deterministic clustering across modules, customers, exact errors, failure language, and action paths. Signals are operational prompts, not confirmed incidents or root causes.
No confirmed multi-customer incident
The engine looks for repeated evidence while preserving the difference between a pattern signal and a verified incident.
Plan the queue as a constrained system.
Seeded simulation converts ticket effort ranges into repeatable completion bands. Risk, ownership, focus factor, and dependencies shape the recommended sequence.
Generate complete Zendesk-ready packets.
The studio enforces the CD-AI A through G contract, character limits, unverified root-cause language, evidence gates, and minimum-data asks.
Make every decision inspectable and reversible.
Versioned policy controls scoring and pattern thresholds. Checkpoints support undo and redo. Integrity checks hash critical state, detect drift, and preserve audit evidence.
Prepared for static hosting, PWA install, and a future API gateway.
This build remains read-only against external systems. The package includes deployment profiles, cache controls, a local bridge, API contracts, schemas, and a stable adapter boundary.
Audit and Recovery
Every local status change, planning move, import, rollback, export, and integrity check is recorded. Source snapshots can be restored without touching Zendesk.