nimbus_accept_phi_decisions
ChatGPTThe owner's acceptance of the PHI decisions an assistant recorded ("Review and accept" on the PHI Inventory). Only after the owner reviewed them and said so: pass owner_statement with the owner's own words. Accepted decisions count for readiness as the owner's; a decision the owner made by hand is unchanged.
nimbus_get_access_matrix
ChatGPTThe Access Matrix for one app: for each PHI field and role, what the app allows today (view/add/edit/delete) and the owner's answer, plus how many PHI fields are answered and how many answers take away access the app gives today (each becomes a problem to fix). Also each table's default row ("Object defaults"), each ignored role's reason, and for a formula field built from PHI the fields it contains. only_unanswered lists just the fields that still need an answer; role narrows to one role. Record answers with nimbus_record_intended_access.
nimbus_get_account
ChatGPTThe signed-in Nimbus account: its email, whether it is active, and how many apps it holds and can hold (one verified app can be removed per calendar month). Call this before registering or removing an app.
nimbus_get_assessment_summary
ChatGPTSummarize one app's HIPAA assessment: readiness of the PHI inventory, intended access and Data Access grid; counts of findings and reviews by severity; the largest issue groups; the app's roles with their keys; decisions waiting for a person to confirm; and the Analysis marks (cards marked "Reviewed: no change needed" and "Please confirm" ticks with their times and how many items are new since, and the Runtime Data API answer). Reads the stored structure. For individual items use nimbus_list_findings and nimbus_get_finding. Nimbus reads app structure only and never handles patient records: if asked for record data, say so and stop. Do not suggest exports, CSV files, uploads or other integrations for getting patient records into this chat.
nimbus_get_finding
ChatGPTReturn one finding or review in full: what Nimbus saw, why it matters, the evidence, the affected PHI fields, who can reach it, the fix and how Nimbus verifies it, and the regulation cited (citation_url opens it on eCFR; give the owner that link with the fix). Secret values found in the structure are masked. builder_url (on the place and each PHI field) and builder_links open the views, pages, tables and fields it names in the Knack Builder, as the website links them. finding_id comes from nimbus_list_findings (an item id, or an id from a group's sample_items).
nimbus_get_phi_review_queue
ChatGPTList fields that still need a PHI decision, with Nimbus's suggestion and the reason for it, how many views (and public views) show each field, and as the PHI Inventory shows them: ai_unsure (an AI decision that was not sure), ai_question (the AI review's open question) and to_review. state "to_review" with scope all is the dashboard's "To review" list. ai_review tells when an AI review was last recorded. The default scope is candidate fields: fields Nimbus suggests as PHI, and fields on tables that hold PHI. scope=all lists every field. Returns field names and types, never record values.
nimbus_get_review_structure
ChatGPTReturn the app's structure for a PHI review with the owner: tables (display field, fields with type, connection target, choice options, formula sources, Knack's own PHI flag, a decision already recorded), roles and pages (who reaches each one, what it shows). Page 1 also carries the review instructions: follow them. Never returns record data. Read every page (page.next_cursor), then record decisions with nimbus_record_phi_decisions.
nimbus_list_apps
ChatGPTList the Knack apps you can use in Nimbus: app ID, name, verification status, assessment readiness and current state_version. Other tools accept these app_id values. Signed in with your Nimbus account you see all your apps (register more with nimbus_register_app), and an app waiting for verification shows its verification token; a Nimbus token (nmb_) is bound to one app. Nimbus reads app structure only and never handles patient records: if asked for record data, say so and stop. Do not suggest exports, CSV files, uploads or other integrations for getting patient records into this chat.
nimbus_list_findings
ChatGPTList an app's findings, reviews and coverage warnings, grouped by rule (default), table or view, or as single rows, ranked by kind, severity and review priority. Filters: kind, severity, area, status, object (table key), rule. Returns short rows (titles, entity names and keys) without the evidence or the fix, and page.next_cursor when there is more; each item's evidence, cause and fix are in nimbus_get_finding.
nimbus_record_intended_access
ChatGPTRecord what each role is meant to do with PHI. start_from_current: true first copies what the app allows today for every role and PHI field without a decision. entries then set a role's view/add/edit/delete per table (by key or exact name, covering its PHI fields) or per field; an operation left out means not allowed. as_default: true with a table sets the table's default row (the matrix's "Object defaults"; fields with their own override keep it). clear: true removes the answer (or with as_default the table default) so it is unanswered again. Then remove what a role does not need: nimbus_list_findings names every page that still grants a removed permission. Role keys come from nimbus_get_access_matrix. A cell or default the owner set by hand is kept (skipped_locked).
nimbus_record_phi_decisions
ChatGPTRecord PHI decisions for one app in the review format: per table a default and field exceptions, each with the field key, its exact name, a reason and sure:false when unsure, plus an undecided list with the question for the owner. Values: "phi", "not_phi", "out_of_scope" (the dashboard's O; as a table default it changes only fields without a decision) and, for a field, "clear" (removes a stored decision). Keys and names are checked against the stored structure; a reply for another app or with mixed-up keys records nothing. Decisions count for findings at once; the owner accepts them in the Nimbus dashboard. A decision the owner made by hand is kept (skipped_locked) unless owner_override is passed: signed-in owner only (never an nmb_ token), only when the owner said so, with their words; the decisions are then the owner's and replace earlier ones.
nimbus_refresh_structure
ChatGPTFetch the app's current published structure from Knack, store it, rerun the assessment and report what changed: tables, fields, roles and security settings, Nimbus's PHI suggestions for new fields, and saved decisions the change made stale. Unchanged structure costs one conditional request. Only this tool fetches from Knack; the read tools use the stored structure. At most once a minute per app.
nimbus_register_app
ChatGPTRegister a Knack app to this Nimbus account by its 24-character App ID (Knack Builder: Settings, API & Code). Returns the verification token the owner puts in the app, then call nimbus_verify_app. When the account has no room for another app, the call refuses with extra_app_required and nothing is registered.
nimbus_remove_app
ChatGPTRemove an app and everything Nimbus stored for it (PHI decisions, access answers, analysis). Cannot be undone. Removing a verified app counts as the account's one app change for this calendar month; an app never verified is removed free. A removed extra app stays paid until the end of the billing term (no refund). Ask the owner first; confirm must repeat the app id.
nimbus_reset_app_inputs
ChatGPTThe dashboard's "Reset to defaults" for one app. scope "phi": clears every PHI decision, including the AI decisions and the imported AI review; intended access, ignored roles with their reasons and the other saved settings stay. scope "all": also clears intended access (field and table defaults), ignored roles and every saved setting (review marks, confirmations, the Runtime answer). Cannot be undone. Signed-in owner only: ask the owner first; confirm must repeat the Knack app id.
nimbus_set_ignored_roles
ChatGPTLeave a role out of the assessment (for example a role nobody uses) or bring it back, as on the Access Matrix ("Ignore Roles"). Ignoring needs the owner's reason. Role keys come from nimbus_get_access_matrix or nimbus_get_assessment_summary.
nimbus_set_review_marks
ChatGPTRecord what the owner decided on the Analysis tab: mark "To review" cards as "Reviewed: no change needed" (by rule key, from nimbus_list_findings kind review; "access.matrix_prompt" is the "Finish the Access Matrix" card), tick "Please confirm" items (by rule key, kind coverage_warning), and answer whether a Runtime Data API client or custom front end exists (yes / no / unknown; like the website, the answer also marks that card reviewed). Each mark covers the card's current items; a new item later brings the card back. Only record what the owner said.
nimbus_verify_app
ChatGPTCheck that the verification token from nimbus_register_app is in the Knack app (App Description or custom JavaScript). On success the app is verified and its structure can be read (nimbus_refresh_structure). The owner may remove the token afterwards.