> For the complete documentation index, see [llms.txt](https://help.nightfall.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.nightfall.ai/data-exfiltration-prevention/exfiltration_endpoint/policies/recommended-policy-configurations.md). # Recommended Policy Configurations These policies protect specific kinds of data wherever they go. They sit on top of the twelve [Recommended Human Exfil Policies](/data-exfiltration-prevention/exfiltration_endpoint/policies/recommended-human-exfil-policies.md), which protect by destination. Build that core set first, then add the policies here that match the data you hold. | # | Policy | Triggers | Action | | - | ------------------------------- | ----------------------------------------------------------------------- | --------------------------------- | | 1 | Credentials and secrets | Browser uploads, paste, desktop app | Block from day one | | 2 | Source code | 2a: Git push. 2b: Browser uploads and cloud sync | 2a: Alert. 2b: Block | | 3 | Regulated data (PHI and PCI) | Browser uploads, paste, desktop app, cloud sync, removable media, print | Monitor, then Block with override | | 4 | Confidential business documents | Browser uploads, cloud sync, desktop app | Monitor, then Block with override | | 5 | Departing employees | Browser uploads, paste, cloud sync, removable media, print | Monitor, then Block | For Claude Code, Cursor, VS Code Copilot, and Claude Enterprise, add the matching agent policies on [AI agent and Inference Hooks policies](/data-exfiltration-prevention/exfiltration_endpoint/policies/recommended-policy-configurations/ai-agent-and-inference-hooks-policies.md). Endpoint triggers do not see what a coding agent does. *** ### Before you start **Agent versions** | Trigger | macOS agent | Windows agent | | -------------------------------------------- | --------------- | ------------------------------------------------------------------------------------ | | Browser uploads, cloud sync, removable media | 1.2.10 or later | 1.4.9 or later | | Clipboard paste | 1.2.11 or later | 1.4.9 or later | | Desktop app | 1.2.11 or later | 1.4.9 or later (Outlook, Teams, WhatsApp). 1.4.11 or later for Microsoft 365 Copilot | | Print | 1.2.11 or later | 1.4.11 or later | | Git push | 1.2.10 or later | Supported | **Setup** * Nightfall browser extension deployed to managed browsers. * Directory sync configured, for scoping by user group. * Collections from the core set: **Sanctioned Apps**, **Approved Git Destinations**, and **Corporate Domains**. See [Recommended Human Exfil Policies](/data-exfiltration-prevention/exfiltration_endpoint/policies/recommended-human-exfil-policies.md). **Actions on every policy** * **Admin alert** to the team that owns the data. * **End-user notification** on every Block, saying what was stopped and who to contact. * **Override with justification** where a legitimate business need can exist (Recommendations 3 and 4). Never on credentials. *** ### Recommendation 1: Credentials and secrets **Use case:** Block API keys, passwords, and private keys from being uploaded or pasted to any destination, including sanctioned ones. Secrets belong in a secrets manager, not in a chat, a ticket, or an AI tool. **Scope** * **Users:** All users. If you stage the rollout, start with engineering and DevOps. **Triggers** (one policy per trigger) | Trigger | Setting | | --------------- | ---------------------------------------------------------- | | Browser uploads | Any domain | | Clipboard paste | Any destination | | Desktop app | Uploads and paste. Covers messaging apps and email clients | **Detection rules** * **API keys and secrets:** AWS, Azure, Google, Stripe, Okta, Slack, GitHub, and 50+ service-specific formats * **Passwords and credentials:** username and password patterns in text and code * **Cryptographic keys:** RSA and EC private keys (PEM-encoded) **Actions** * **Automated action:** Block. Credential detectors are specific enough to block from day one. * **End-user notification:** *"A secret or API key was detected and cannot be shared. If it was already exposed, rotate it now."* *** ### Recommendation 2: Source code **Use case:** Detect source code pushed to personal or unapproved repositories, and block source code from being uploaded or synced to personal and unsanctioned destinations. Build this as separate policies. Nightfall can block an upload, but it can only alert on a git push. #### Policy 2a: Git push to unapproved repositories (alert) This is the same policy as core policy 7. If you already have it, skip to 2b. * **Users:** Engineering and DevOps, or all users with Git installed. * **Trigger:** Git push to destinations not in **Approved Git Destinations** (for example `github.com/your-org/*` and `gitlab.yourcompany.com/*`). Paths are supported, so you can approve your organization without approving all of `github.com`. * **Detection rules:** None. Content scanning is not available for this trigger. * **Actions:** Admin alert. The push completes. Nightfall records the user, device, repository, remotes, and destination URL. There is no end-user notification. See [Git Push Monitoring](/data-exfiltration-prevention/exfiltration_endpoint/policies/trigger/git-push-monitoring.md) for how matching works. #### Policy 2b: Source code uploads (block) * **Users:** Engineering and DevOps. * **Triggers** (one policy per trigger): | Trigger | Setting | | --------------- | ----------------------------------------------------------------------------------------------------------------------------------- | | Browser uploads | Domain in **Code repositories**, **GenAI**, and **Agent builders**, excluding **Approved Git Destinations** and **Sanctioned Apps** | | Cloud syncing | Personal sync folders | * **Detection rules:** Source code file classifier. Optionally add custom regex for internal project identifiers, copyright headers, or proprietary module names. * **Actions:** Block, with an admin alert and the end-user notification *"Company source code cannot be uploaded to personal or unapproved destinations."* To scan code committed to your GitHub organizations, see [Recommended Policies for GitHub](https://help.nightfall.ai/sensitive-data-protection/github/policies/recommended-policies-for-github). *** ### Recommendation 3: Regulated data (PHI and PCI) **Use case:** Stop health information and payment card data from leaving through any channel except the systems approved to hold it. This supports HIPAA and PCI DSS programs. Build one policy per data type, so alerts reach the right team. | | PHI | PCI | | ------------------- | -------------------------------------------------------------- | ---------------------------------------------------------------------------------------- | | **Detection rules** | PHI, plus PII (names, SSNs, dates of birth) | PCI (card numbers, routing numbers, IBAN, SWIFT), plus the financial document classifier | | **Users** | All users, or clinical, operations, and data teams for a pilot | Finance, accounting, and billing first, then all users | | **Alert route** | Compliance and security | Security and compliance | **Triggers** (one policy per trigger) | Trigger | Setting | | ----------------------------------- | ---------------------------------------------------- | | Browser uploads and clipboard paste | Destinations not in **Sanctioned Apps** | | Desktop app | Uploads and paste, with exceptions for approved apps | | Cloud syncing | Personal sync folders | | Removable media | Enabled | | Print | Enabled | **Why not every destination:** Clinical and finance staff move this data into approved systems all day. A policy that blocks every destination stops that work, and people route around it. **Actions** * **Automated action:** Monitor for two weeks, then Block. * **Override with justification:** On. Review justifications weekly. They form the audit trail. * **End-user notification:** *"This content contains protected health information (or payment card data) and can only be shared through approved systems. Contact your compliance team for help."* Git push is not included. It does not scan content, so it cannot match these detectors. *** ### Recommendation 4: Confidential business documents **Use case:** Stop M\&A materials, legal contracts, HR records, and internal financial reports from leaving through unsanctioned destinations. **Scope** * **Users:** Leadership, finance, legal, HR, and strategy teams, scoped by directory group. * **Asset origin (optional):** On browser uploads, set **Source in** your corporate collections with **Corporate accounts only** on. This follows files that came from your corporate Drive or SharePoint, wherever they go. **Triggers** (one policy per trigger) | Trigger | Setting | | --------------- | ------------------------------------------ | | Browser uploads | Destinations not in **Sanctioned Apps** | | Cloud syncing | Personal sync folders | | Desktop app | Uploads, with exceptions for approved apps | **Detection rules** * **File classifiers:** M\&A, legal, financial, HR, and regulatory documents * **Custom keywords (optional):** internal project codenames, product names, or deal names **Actions** * **Automated action:** Monitor for two weeks, then Block. * **Override with justification:** On. Many document transfers have a legitimate reason, and the justification is your record. *** ### Recommendation 5: Departing employees **Use case:** Record everything a departing employee moves out of corporate accounts during their notice period, and stop bulk transfers to devices and personal storage. **Scope** * **Users:** A **Departing users** directory group. Add people when notice is given and remove them after their last day. **Policies** | Policy | Triggers | Detection rules | Action | | ------------------------------ | --------------------------------------------------------------------------------------------------------------------------- | ----------------------------------- | ------- | | Record corporate data movement | Browser uploads and paste with **Source in** your corporate collections and **Corporate accounts only** on. Any destination | None (content scanning off) | Monitor | | Stop bulk transfers | Removable media, cloud syncing, print | Standard set from the core policies | Block | The first policy has content scanning off on purpose. It records every movement out of a corporate account, regardless of content, so you have a complete record for the notice period. Do not set it to Block, because with content scanning off that would stop all of that person's uploads. To review what someone moved, search their activity in [Forensic Search](/data-exfiltration-prevention/forensic-search/forensics-search.md). *** ### Where the earlier recommendations went This page previously listed nine recommendations. Several duplicated the core set, so they now live there. | Earlier recommendation | Now | | ---------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | 1. Block sensitive data to external AI tools | Core policies 1, 4, 9, and 10. For Claude Enterprise, see [AI agent and Inference Hooks policies](/data-exfiltration-prevention/exfiltration_endpoint/policies/recommended-policy-configurations/ai-agent-and-inference-hooks-policies.md) | | 2. Personal cloud storage | Core policies 2 and 8 | | 3. Credential and API key leaks | Recommendation 1 | | 4. Source code | Recommendation 2 | | 5. PHI (HIPAA) | Recommendation 3 | | 6. Removable media | Core policy 11 | | 7. Financial data (PCI DSS) | Recommendation 3 | | 8. Corporate IP | Recommendation 4 | | 9. Paste to personal accounts and risky destinations | Core policies 4, 5, and 6 | *** ### Tuning * **Roll out the same way as the core set.** Monitor, then notify, then block. Credentials are the only exception. See [Building the set](/data-exfiltration-prevention/exfiltration_endpoint/policies/recommended-human-exfil-policies.md). * **Keep Sanctioned Apps current.** Recommendations 2 to 4 work by exception, so a missing sanctioned app becomes a false positive. Use [App Intelligence](/data-exfiltration-prevention/app-intelligence/app-intelligence.md) to find apps in use and decide which belong in the collection. * **Add custom detectors** for data specific to your business, such as customer IDs or internal document markings. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://help.nightfall.ai/data-exfiltration-prevention/exfiltration_endpoint/policies/recommended-policy-configurations.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.