Data We Collect

Transparency is a core F7 value. This page provides a complete breakdown of what the F7 agent captures, what it never captures, and why each data point exists.

What We Capture

Data Category	Examples	Why We Capture It
Application focus	App name, app category, foreground/background transitions	Understand which tools people use and how they divide their time
Input activity	Click count, keystroke count, scroll events, idle periods	Measure engagement levels without recording what was typed
Network metadata	Destination domain, port, detected AI provider	Identify which AI tools are being used and how frequently
Session structure	Session start/end, duration, focus time, context switches	Understand work patterns, deep work vs. fragmented time
AI interaction metadata	AI provider name, turn count, request/response sizes, time-to-first-response	Measure depth and sophistication of AI usage
Git metadata	Commit count, files changed count, insertions/deletions	Correlate AI usage with development output (via webhook, not agent)

Employer-Provided Data

In addition to agent-captured metadata, F7 processes organization-provided workforce data such as identity, role, and hierarchy fields when the deploying organization configures those syncs.

Data Category	Examples	Why It's Used
Job metadata	Job title, employment type, cost center, location	Team-level analytics and segmentation
Org hierarchy	Department, division, team, manager	Reporting structure and rollup views
Employment lifecycle	Hire date, termination date, active status	Accurate headcount and tenure analysis
Identity	Email, display name, user principal name	Account linkage and HRIS/IdP sync

Employer-Controlled Data

F7 does not independently collect workforce directory data — it is provided entirely by the deploying organization. The organization decides which supported fields to sync and can disable those syncs at any time.

Third-Party Event Integrations

With your organization's authorization, F7 can ingest event metadata from configured controller integrations. This provides richer analytics without relying solely on network-level observation from the agent.

Integration Category	Implemented Sources	Data Retrieved
Source control	GitHub, GitLab, Bitbucket	Commit, pull request, branch, and review event metadata
Issue/project management	Jira, Linear, Asana	Ticket and sprint event metadata
CI/CD	Jenkins, GitHub Actions, GitLab CI, CircleCI	Build, test, and deployment event metadata
Calendar/support/feature flags	Google Calendar, Microsoft Outlook, Zendesk, LaunchDarkly	Meeting, support-ticket, and flag-change event metadata
Custom workflows	Generic webhooks	Customer-defined event metadata

Metadata Only — Never Content

Third-party integrations retrieve event metadata only — never document contents, message text, prompt/response text, code diffs, repository contents, calendar descriptions, support conversation bodies, or file contents. F7 requests only the minimum scopes required for the configured event source.

Opt-In Only

Third-party integrations are not enabled by default. Each integration must be explicitly authorized by an organization administrator, who grants the specific API credentials and scopes. Integrations can be disconnected at any time.

Important Details

• Keystroke count, not keystrokes. We count how many keys were pressed. We never record which keys or what was typed.
• App name, not window title. For non-work applications, we capture only the app name and category.
• Domain plus AI-provider endpoint path, not full URL. We record the destination domain and, for AI-provider API calls, the endpoint path used for classification. We never capture query parameters, request bodies, response content, or page content.
• Sizes, not content. For AI interactions, we record how large requests and responses were. Never the actual prompt or response text.

What We Never Capture

Category	Guarantee
Prompt and response text	Never captured — not even with advanced features enabled
File contents	Never read — only metadata like file count and type
Email and chat content	Never captured — only time-in-app and domain
Screenshots	Never transmitted off the device — if the optional vision model is enabled, frames are processed locally; current macOS capture streams PNG bytes through stdout and the agent scrubs stale legacy temp files on startup
Clipboard contents	Never captured under any circumstances
Passwords or credentials	Never accessed or stored
Browsing history	Only the active destination domain is captured, with AI-provider API endpoint paths used for classification — never full URLs, query parameters, request bodies, response content, or page content
Personal app activity	Apps on the exclusion list generate zero telemetry
Individual keystrokes	Only aggregate counts — never what was typed

Where Data Lives

Location	What's Stored	Encryption
Employee's device	Raw observations + classified sessions (temporary)	AES-256 encrypted local database
F7 Controller	Structured metadata + computed scores	Encrypted at rest, tenant-isolated
Dashboards	Rendered analytics (not stored separately)	Served over HTTPS

Data Retention

Data Type	Default Retention	Configurable?
Raw telemetry	90 days	Yes — per organization
Daily/weekly rollups	12 months	Yes — per organization
Audit logs	24 months	Minimum 24 months
Agent local data	Synced then pruned	Automatic

Organizations can configure retention periods. When data is deleted, it is permanently removed from all storage — not archived.

Right to Erasure

Any employee's data can be fully erased on request:

1. The organization issues an erasure command through the admin interface
2. The controller deletes all records for that individual
3. The agent receives a remote command to wipe all local data
4. An audit trail of the erasure itself is retained (for compliance — without the original data)

---

For Procurement Teams

This page is designed to answer common vendor security questionnaire items about data collection scope. For encryption and access control details, see the Security Overview.