privacy-logging-and-telemetry-audit

# Privacy Logging And Telemetry Audit Audit SimpleChat logging, telemetry, activity records, plugin logs, traces, diagnostics, exports, and error handling for privacy and sensitive-data exposure. Use the repository guardrails in [.github/instructions/santize_settings_for_frontend_routes.instructions.md](../../instructions/santize_settings_for_frontend_routes.instructions.md), [.github/instructions/broken-access-control-prevention.instructions.md](../../instructions/broken-access-control-prevention.instructions.md), [.github/instructions/python-lang.instructions.md](../../instructions/python-lang.instructions.md), and [.github/copilot-instructions.md](../../copilot-instructions.md). ## Operating Rules - Work in the SimpleChat repository root. - Treat user prompts, chat messages, document text, uploaded content, generated outputs, profile fields, emails, IDs, headers, tokens, settings, and tool responses as sensitive unless explicitly intended for that log audience. - Do not revert unrelated user changes. - If the user asks for `scan only` or `stop after plan`, do not edit files. - If fixing findings, preserve useful diagnostics while minimizing sensitive payloads. - Prefer `log_event(...)` from `functions_appinsights.py` and existing activity/plugin logging helpers over ad hoc logging. ## Baseline Discovery 1. Confirm the repo root, current branch, and concise `git status --short`. 2. Inventory logging and telemetry calls: ```powershell rg -n "log_event\(|logging\.|logger\.|print\(|traceback|AppInsights|Application Insights|activity_log|activity_logs|plugin_log|plugin_logging|thought|telemetry|diagnostic|debug|error|exception" application functional_tests scripts deployers ``` 3. Search for likely sensitive fields near logs, exports, and diagnostics: ```powershell rg -n -i "prompt|message|content|document|chunk|summary|citation|email|user_id|object_id|authorization|cookie|token|secret|password|api[_-]?key|connection[_ -]?string|headers|settings|request\.|response\." application/single_app ``` 4. Search export and admin-view surfaces: ```powershell rg -n "export|download|csv|json|xlsx|docx|pdf|report|activity|control_center|admin|audit|history|trace" application/single_app docs functional_tests ``` 5. Search for broad serialization of request, response, settings, or exception objects: ```powershell rg -n "vars\(|__dict__|json\.dumps\(|asdict\(|model_dump\(|dict\(|request\.headers|request\.cookies|request\.environ|os\.environ|str\(.*exception|repr\(" application/single_app ``` ## Manual Audit Checklist Review these areas: - `log_event(...)` custom properties and messages for secrets, tokens, headers, cookies, settings, connection strings, prompts, document text, or full tool responses. - Activity logs and admin views for user-visible privacy boundaries, searchability, retention expectations, and unnecessary PII. - Plugin invocation logs, thought records, workflow activity, agent citations, and generated artifacts that may store raw external responses or sensitive context. - Exception handlers and error responses that expose stack traces, request payloads, service errors, internal endpoints, IDs, or secrets. - Export flows that package chat history, logs, citations, metadata, activity, or diagnostics with more data than the user/admin should receive. - Debug flags, environment-controlled diagnostics, and support endpoints that can be enabled in production. - PII fields such as display names, emails, Entra IDs, group memberships, profile images, IP addresses, document names, and workspace names. - Whether logs include enough correlation IDs and coarse event names after redaction to remain useful. ## Triage And Plan Group findings before editing: - `Critical`: Secrets, tokens, connection strings, authorization headers, or raw private content are logged or exported to a broad audience. - `Important`: Sensitive user, document, prompt, or tool data is logged by default without clear need or masking. - `Moderate`: Debug diagnostics, admin views, error responses, or exports are overbroad but role-restricted or environment-limited. - `Low`: False positives, placeholders, reviewed diagnostic fields, or coarse metadata that is necessary and proportionate. For each finding, record: - File and logging/export surface. - Sensitive source. - Audience and storage sink. - Existing redaction or role boundary. - Missing minimization, masking, or authorization. - Realistic impact. - Remediation approach. - Minimum regression test. ## Remediation Patterns Use these fixes by default: - Log event names, IDs, counts, durations, statuses, and coarse categories instead of raw payloads. - Redact secrets, headers, cookies, tokens, connection strings, and SAS URLs before logging. - Truncate long prompt, document, tool, or service-response previews and avoid logging full content by default. - Use stable correlation IDs rather than full user/profile payloads when possible. - Keep admin exports role-protected and scoped to explicit filters. - Sanitize settings before frontend diagnostics unless it is an admin route that deliberately requires raw settings. - Add tests for redaction helpers, log payload construction, export field filtering, or error response shape. ## Verification After fixing, run the narrowest reliable checks: - `python -m py_compile <changed-python-files>`. - Relevant functional tests under `functional_tests/`. - Targeted `rg` searches from the discovery section for changed areas. - `python scripts/check_broken_access_control.py --full-file <changed-python-files>` when log/export access control changed. - `git -c core.whitespace=blank-at-eol,blank-at-eof,space-before-tab,cr-at-eol diff --check`. If a check cannot run locally, explain why and include the remaining risk. ## Output Expectations Return findings first, ordered by severity. Include log and export surfaces reviewed, files changed, redaction/minimization patterns used, validation commands run, skipped checks, false positives, and remaining risks.