Data and access · Private pilot
We read. We never change anything.
The product is being built around read-only collection, explicit assessment scope, and evidence you can inspect.
Access boundaries
Plimie will only read provider data needed for enabled checks. It will not change WAF policies, cache settings, DNS, or infrastructure. Customers remain responsible for reviewing and applying recommended actions.
Credentials
Customer credentials are encrypted before storage with AES-256-GCM, a separate encryption secret, and workspace/connection-bound authentication. Tokens are not saved in browser storage or returned by the API. Disconnect removes the stored credential and cancels active scans. An already-started provider read can finish, but cancelled scans cannot append results. Revoke the token separately in the account where you created it.
One-time resource checks use credentials only for the selected request and do not store them. An owner may separately choose to save the normalized result, including linked cache-rule identifiers/action summaries, route-linked origin-group and origin identifiers with enabled state, probe and balancing settings, selected-route security-policy association status, linked policy identifiers and domain counts, linked policy enabled/mode status and configured bot-rule-set presence, selected distribution origin and failover-group identifiers with configured status codes and behavior-target counts, selected backend group identifiers and balancing settings with linked health-check timing/thresholds and aggregate current health-state counts, selected route-policy map identifiers, generic branch positions and modes, configured frontend proxy and forwarding-rule identifiers, link counts and routing scheme, backend cache mode, configured TTLs and cache-key inclusion switches, verified profile tier and dated cache-pricing context, an optional caller-account service-cost observation (account identifier, positive total and currency or unknown, estimated flag and UTC window), a partial project-level CDN-labelled SKU cost observation (project identifier, positive total and currency or unknown, and UTC window), a profile-attributed pre-tax actual cost observation (positive total and currency, or unknown) with a UTC window, and an aggregated seven-day viewer-request metric, cache-result breakdown, daily cache-hit percentage and additional-metrics subscription state, policy-level allowed/blocked or preview-rule match count, selected top-level rule action count, or client access-log observation or recent grouped WAF rule samples for a selected resource; workspace members can view it, and it appears in owner exports. Cache-key parameter/header/cookie names, route host/path match values, origin hostnames and paths, behavior path patterns, probe paths, health-check host/path values, endpoint IP addresses and instance identifiers and raw provider responses and billing rows are not saved. These observations do not establish effective route policy, cache suitability, rule-level effectiveness, complete or feature-specific billing, or account-wide coverage.
Evidence and limitations
Configuration, analytics, entitlements, cost, and customer confirmations are distinct evidence sources. Missing permissions and unsupported plans remain visible gaps. An inferred model score cannot create a recommendation by itself.
Subscription evidence
With Billing Read access, scans retain selected-zone subscription terms and recognized account-level add-ons: product, status, reported price, currency, period and management flags. Account prices are not assigned to individual zones. Workspace members can view this evidence; exports contain it. Payment-card details, billing addresses and invoice documents are not retained by these collectors. Missing or conflicting evidence remains unknown.
AI processing
Plimie does not send customer data to third-party AI providers. Optional statistical forecasting uses only aggregated scan metrics, processed by our hosting provider, and you can turn it off in workspace settings. Scans record that inferred signals are unavailable and continue with deterministic rules. Plimie does not send customer data to an external AI model. Any later model input must be minimized and stripped of credentials, raw responses, IP addresses, cookies, and full URLs.
An optional owner-initiated log check processes bounded request records transiently to count matches by linked policy, rule priority, action, result and preview/enforcement state. Only normalized counts and coverage are saved on explicit opt-in. Temporary log credentials and raw visitor IPs, URLs, headers and request rows are not saved. Positive counts are lower bounds; missing logs do not prove unused rules.
An optional recent rule-sample check transiently processes bounded request samples, which can include visitor IPs, URLs and headers. Only grouped rule/action sample counts and coverage are saved on owner opt-in; raw requests are discarded. Samples may span multiple protected resources and are not total traffic.
An optional selected-policy WAF diagnostic query returns grouped rule-event counts for verified linked resources; it does not return raw visitor request rows. Owners can save the normalized rule name, action, mode and count for members and exports. Multiple rule events can come from one request, and missing diagnostic logs remain unknown.
Scan and report notifications
In-app notices and weekly reports are scoped to your workspace membership. Owners can separately opt in to email their own sign-in address when scans finish or weekly reports are ready. Both email preferences start off. Email contains a generic notice and authenticated link, not evidence, results, or credentials. Opt-out stops pending delivery, but messages already in flight cannot be recalled. Deleting a workspace removes its report, notice, and preference records, not copies already held by email providers or recipient mailboxes.
Retention and deletion
This release is for the invited owner to validate the pilot. Evidence and feedback persist in the application database until the owner deletes the workspace; disconnecting a connection does not delete them. No approved backup or timed-retention duration exists yet. Owners can export their workspace data without provider credentials or sign-in data. Owners can delete a workspace from the workspace page after confirmation; that cancels in-flight scans, wipes stored credentials, and removes the workspace’s records. Read our Privacy Policy and Terms of Service, operated by Zeroteam LLC, Texas, United States. For support or privacy requests, email contact@plimie.com. The sample assessment is fictional; its interactive feedback is held only in the current page and resets on reload.
Workspace deletion removes the application’s database records and durably queues removal of retained scan-processing state. Cleanup retries keep only instance IDs and operational metadata until the provider acknowledges deletion or absence. Cleanup is asynchronous, not immediate. Database recovery copies are separate; their retention policy and the live cleanup lifecycle must be verified before admitting external customers.