Investigation Management

12 entities and 60 agent-callable operations in investigation. Each operation is addressed as {mega_tool}/investigation/{entity}/{operation} and carries the same permission check over MCP and REST.

analysis_request

OperationMega-toolWhat it does
listentity_crudList analysis_requests with filtering and pagination
readentity_crudGet detailed analysis request information
searchentity_crudSearch analysis_requests with fuzzy matching

deidentification

OperationMega-toolWhat it does
approvalsreport_queryExport approvals recorded for a DFCS investigation: who approved which scope under which method version and date granularity, the sample counts they approved…
approve_exportsmart_actionApprove the export of a de-identified scope.
countsreport_queryWhat the de-identification pass actually did, by type: entities detected, values restored by the allowlist, detections dropped below threshold, and the…
exportsmart_actionWrite the approved de-identified records of a scope to a JSONL file on the host.
listreport_queryDe-identified records you may see, newest month first: id, record key, status, review decision, stratum and token counts.
review_decisionsmart_actionRecord a human accept or reject decision on ONE de-identified record, after reading its text in the review tab.
revoke_exportsmart_actionRevoke a live export approval.
samplereport_queryThe human review sample for one de-identification run — the 40 records drawn by stratum (records carrying child/relative pseudonyms, records the residual…
set_profilesmart_actionSet this organization's de-identification profile: which agency and provider domains and names are correspondents rather than patients (and so are restored…
startsmart_actionStart a de-identification pass over a DFCS (also written DFACS) investigation: segment every held record, scrub it with AWS Comprehend Medical DetectPHI plus…
statusreport_queryState of one de-identification pass over a DFCS (also written DFACS) investigation: the run row (state, stop reason, spend against ceiling), how many text…

investigation

OperationMega-toolWhat it does
createentity_crudCreate a new investigation
deleteentity_crudDelete an investigation
listentity_crudList investigations with filtering and pagination
listsmart_actionList investigations matching the supplied filters.
publishsmart_actionPublish an investigation report from a Typst source.
readentity_crudGet detailed investigation information
searchentity_crudSearch investigations by text query
updateentity_crudUpdate an investigation

investigation_finding

OperationMega-toolWhat it does
createentity_crudCreate a new investigation finding
deleteentity_crudDelete an investigation finding
listentity_crudList investigation findings with filtering
readentity_crudGet detailed finding information
searchentity_crudSearch investigation findings
updateentity_crudUpdate a finding

investigation_folder

OperationMega-toolWhat it does
createentity_crudCreate a new investigation folder
deleteentity_crudDelete an investigation folder
listentity_crudList investigation folders
readentity_crudGet folder details
searchentity_crudSearch investigation folders by name
updateentity_crudUpdate a folder

investigation_folder_assignment

OperationMega-toolWhat it does
createentity_crudCreate a new investigation folder assignment
listentity_crudList investigation_folder_assignments with filtering and pagination
readentity_crudGet detailed investigation folder assignment information

investigation_phase

OperationMega-toolWhat it does
createentity_crudCreate a new phase
deleteentity_crudDelete a phase
listentity_crudList investigation phases
readentity_crudGet phase details
searchentity_crudSearch investigation phases
updateentity_crudUpdate a phase

investigation_run

OperationMega-toolWhat it does
calibrationreport_queryThe model-selection evidence for this investigation: per-rung F1 / recall on dfacs_related, macro-F1 on interaction type, exact match on referral outcome…
cancelsmart_actionRequest cancellation of a running DFCS investigation stage.
countsreport_queryThe monthly series behind the DFCS timeline, keyed YYYY-MM: the referral funnel (inquiries → referrals → admissions, with declines split by side) over…
dry_runsmart_actionProject what the next DFCS stage would cost on this run — rows remaining, per-item list price, projected USD and whether it fits the ceiling.
listreport_queryEvery DFCS investigation run you may see, newest first: kind, state, stop reason, spend against ceiling and the models chosen.
startsmart_actionStart one stage of the DFCS (also written DFACS) relationship investigation pipeline on an existing restricted investigation: stage0 (metadata census, free)…
statusreport_queryState of one DFCS (also written DFACS) investigation run — stage, stop reason, spend against ceiling, chosen models, per-model usage and the stage counters…

investigation_share

OperationMega-toolWhat it does
listentity_crudList investigation_shares with filtering and pagination
readentity_crudGet detailed investigation share information

investigation_workstream

OperationMega-toolWhat it does
createentity_crudCreate a new workstream
deleteentity_crudDelete a workstream
listentity_crudList investigation workstreams
readentity_crudGet workstream details
searchentity_crudSearch workstreams
updateentity_crudUpdate a workstream

problem_list

OperationMega-toolWhat it does
problem_list_summarysmart_actionSummarize a patient's active problem list (EHR investigation context).

report

OperationMega-toolWhat it does
publish_investigationsmart_actionPublish an investigation report from a Typst source file.

Parameter schemas and permissions

Parameter schemas and permission codenames are in the full reference — sign in — or call prism_introspect(dimension="capability_search", query="…") with your key.

In the full reference this application appears under entity_crud, report_query, smart_action.