Dentrix SQL and DEXIS Explorer
Use Agent View to inspect Dentrix SQL records, audit DEXIS image metadata, and run deterministic non-standard claim attachment workflows without leaving the platform.
Open the tools
- Go to
/app/agents. - Open an agent and enter Agent View.
- Use
SQL Explorerfor direct SQL query execution on the worker. - Use
DEXIS Imagesfor patient-to-image inventory lookup, preview conversion, and tooth composites. - Use
Claim Workflowfor non-standard claimScope -> Validate -> Gather -> Review -> Queue.
Use SQL Explorer
Run deterministic SQL discovery and diagnostics:
- Select a template (schema scan, patient lookup, image-table checks).
- Configure or switch connection profiles.
- Execute queries and review tabular or raw output.
- Copy result data as TSV or JSON for triage handoff.
Operator guidance:
- Keep queries bounded with
TOP Nand explicitWHEREclauses. - Treat this panel as investigation tooling; do not use write queries in production workflows.
Use DEXIS Images
Inspect DEXIS inventory from Dentrix identifiers:
- Search by patient name, chart number,
PATID, orPATIMAGEID. - Resolve patient in
DDB_PAT, then derive the digit-sharded DEXIS path. - Parse
DEXIS.datand display image inventory with date, type, and tooth mapping. - Filter by image type, tooth, and date range.
- Use
Gridview for thumbnail browsing. - Use
Tableview for sortable metadata. - Use
Fieldsview for raw field-level diagnostics fromDEXIS.dat. - Use
Scopeview to classify procedures asStandardorNon-Standardand generate attachment requirement signals per procedure (XR,Perio,Narr,Pros) with image and provider-note availability indicators. - Optionally apply a service-date filter (
YYYY-MM-DD) inScopeview to scope procedure analysis to one date; without a date filter, the panel analyzes the most recent 90 days. - Use
Generate Composite Imageto request ADA tooth numbers (1-32) and create one labeled PNG from the most recent image per requested tooth.
When image previews are needed, the panel calls /api/agents/{id}/dexis-images and polls /api/agents/{id}/dexis-images/status to convert .dex TIFF payloads into JPEG assets.
When tooth composites are needed, the panel calls /api/agents/{id}/dexis-composite and polls /api/agents/{id}/dexis-composite/status. The worker resolves the patient through Dentrix SQL, decodes tooth tags from DEXIS.dat field [10], assembles a labeled PNG, uploads it to S3, and returns composite_key for /api/assets/view?key=....
Use Claim Workflow
Run the non-standard attachment readiness flow from Agent View:
Scope: scan a date range for patients with non-standard ADA codes.Validate: select one patient/date claim bundle and confirm code mix.Gather: build a virtual file from Dentrix procedures, recent clinical notes, tooth-linked DEXIS inventory, prior tooth-procedure history, and evidence-of-need phrase matches.Review: confirm readiness signals for required evidence (XR,Perio,Narr,Pros), select per-procedure actions, and finalize an overall decision (approve,hold,skip).Queue: track persisted work orders by status, inspect readiness/decision detail, review execution timeline, and retry or delete failed records.
Implementation details:
ScopeandGatherrun deterministic worker snippets through/api/agents/{id}/run-snippetand poll/api/agents/{id}/run-snippet/status.Scopepatient labels are emitted from Dentrix asLASTNAME, FIRSTNAME.Gatherstarts with procedure-linked teeth from the selected claim.- If procedures provide no tooth linkage,
Gatherfalls back to note-derived teeth only when note date equals the selected claim service date. - If tooth linkage is still missing,
Gatherfalls back to same-day DEXIS images as a last resort and matches those images to unresolved procedures. Gatherclassifies prosthesis context per tooth from prior ledger history asinitialorreplacement, and extracts prep-date hints from recent crown/prosthesis-related procedures.Gathertags notes with provider-authorship and evidence phrases (fracture,caries,needs crown,open margin, etc.) to support deterministic narrative checks.Reviewcomputes deterministic readiness flags (pass,warn,fail) and score across X-ray, narrative, perio, and prosthesis checks.- X-ray readiness aligns to
Y(recent),!(found but older than 6 months), orN(missing); a generated composite image is treated as available evidence for readiness display. Reviewmarks narratives asYwhen provider-authored evidence is present,!for partial conditions (for example DOS note without evidence phrase), andNwhen no usable note is found.Reviewpersists work-order data locally per agent (mimic_wo_*keys) so Queue state survives panel refreshes.Queuestatus model isdraft,approved,executing,completed,failed,held; execution writes step-level detail and screenshot metadata into the same work-order record.- Composite generation reuses
/api/agents/{id}/dexis-compositeand/api/agents/{id}/dexis-composite/status, withpatientLookupset to the selectedPATIMAGEIDstring for deterministic patient resolution. - Outputs are marker-delimited JSON payloads for stable UI parsing and operator review (
__CW_SCOPE_*,__CW_GATHER_*,__CW_EXEC_*).
Operating model
This surface follows Mimic’s RPA-First, LLM-Fallback philosophy:
- Metadata resolution and parsing are deterministic.
- Worker-side conversion is explicit and auditable.
- No agentic inference is required for Dentrix/DEXIS inventory checks or claim attachment readiness classification.