Dataset restrictions are in public beta on the Honeycomb Enterprise plan.
- Protect regulated data: Limit who can read datasets that carry payment, health, or personal data, while the rest of your team keeps working with everything else.
- Keep team-owned data with its team: Let each team read its own sensitive datasets without hiding the rest of your telemetry.
- Protect confidential projects: Limit unreleased or confidential work to the people on the project.
- Separate client data in shared environments: Give each group access only to the client datasets it works on.
- Stop maintaining workarounds: Replace a separate team or duplicate, masked datasets with restrictions on the data you already send.
- Show auditors how access changed: Use the Activity Log to show who changed restrictions and group access, and when.
How dataset restrictions work
Dataset restrictions combine restricted datasets, restricted environments, and user groups to decide who can read what.Restricted datasets
Restricting a dataset hides its contents while keeping it discoverable. Only people with access can read its events, fields, and query results. The dataset’s name stays visible to everyone on the team, so people know the data exists and can ask a Team Owner for access. Everyone on the team keeps full access to unrestricted datasets. To restrict a dataset, visit Restrict Access to Datasets and Environments. To learn what people without access can read and change, visit What Restricted Users Can Do.Restricted environments
Restricting an environment protects all of its data at once. The restriction covers every dataset in the environment, including datasets created later. When you restrict an environment, Honeycomb removes the individual restrictions from its datasets. If you later remove the environment’s restriction, those datasets become unrestricted. In an unrestricted environment, new datasets start unrestricted. You can restrict regular environments and S3 Archive environments. Restrictions don’t apply to system environments, so everyone keeps access to them. The exception is the Activity Log environment, which only people who can read every restricted dataset on the team can query.User groups
User groups give specific team members access to restricted data. For each environment, a Team Owner chooses whether a group can read all of the environment’s datasets, only its unrestricted datasets, or a custom set of restricted datasets. A person can belong to more than one group. Their access combines the access of every group they belong to. To create groups and grant access, visit Manage User Groups.Roles
Restrictions apply to Team Members and Read-Only users; Team Owners always keep full access. Team Owners can read every dataset, whether or not they belong to a group. Restrictions also limit some changes a role otherwise allows. For details, visit What Restricted Users Can Do. To learn more about roles, visit Manage Team Permissions.API keys
Configuration Keys read data without a person attached, so dataset restrictions don’t apply to them.- A Configuration Key can read and write every dataset its permissions cover, including restricted datasets.
- A Management Key with the
groups:writescope can grant any user group access to any individually restricted dataset on your team. To give a group access to a restricted environment, use the Honeycomb UI. - A Management Key with the
environments:writescope can restrict and unrestrict datasets through the Honeycomb API. To unrestrict a dataset that a user group has access to, the key also needs thegroups:writescope. To restrict an environment, use the Honeycomb UI. To learn more, visit Manage Dataset Restrictions with the API. - Honeycomb hides an environment’s API keys from people who can’t read every dataset in that environment. Team-wide keys and the API Keys tab in Team Settings require access to every dataset on the team.
AI features
AI features act for a person, so they read data with that person’s access.- Honeycomb MCP and Canvas run queries with the access of the person using them. For exceptions, refer to Limits and known limitations.
- Auto-investigations run as your team’s agent, not as a person, so they read only unrestricted datasets.
Notifications and Slack link previews
Notifications and Slack link previews can reach people Honeycomb can’t check, so plan where data from restricted datasets goes.Notifications
Recipients of email, Slack, PagerDuty, Microsoft Teams, and webhook notifications may not be Honeycomb users, so Honeycomb can’t check their access before sending.- Email, Slack, and Microsoft Teams notifications contain links, not graph images.
- Slack Trigger notifications include each result’s value and group-by values in the message text, whether or not the data is restricted.
- PagerDuty alerts leave out the graph image when the alert covers restricted data.
- Webhook notifications always include result values, whether or not the data is restricted.
- When a Trigger uses restricted data, Honeycomb reminds its author that restricted data may be visible in notifications.
- Links in notifications lead back to Honeycomb, which checks access when someone opens them.
Slack link previews
Anyone in a Slack channel can view a link preview, so Honeycomb leaves out graphs that include restricted data.- Links to queries, board query results, traces, and the Service Map unfurl without a graph image when the content includes restricted data.
- The preview keeps the title, query description, and link, with a footer that explains why the graph is missing.
- Honeycomb decides by the data, not the viewer, because anyone in the channel might be restricted.
Audit changes
The Activity Log shows who changed access to restricted data. Honeycomb records changes to dataset restrictions, user groups, group members, and group access in the Activity Log.Limits and known limitations
Plan around these limits when you configure dataset restrictions.- A team can have up to 50 user groups.
- Changes to an environment’s restriction setting don’t appear in the Activity Log. Honeycomb plans to add these events.
- Downloading query results as CSV or JSON.
- Canvas downloads of stored query results.
- Canvas investigations that every team member can open. Their tables, summaries, notes, chat, and agent session transcripts can include restricted data. Charts and query results in them stay locked.
- Findings in Canvas, and Slack link previews of Findings.
- The Honeycomb MCP tools
get_query_results,get_query_history,find_queries,list_boards,list_findings, andget_finding. - The Honeycomb MCP tool
run_bubbleupwhen it returns a saved BubbleUp result for an environment-wide query. - The
honeycomb://query-run/...andhoneycomb://trace/...Honeycomb MCP resources. - The
query_result_jsondownload link that the Honeycomb MCP query tools andget_tracereturn.