What is ScreenJournal MCP?

Updated on 14 September 2026

ScreenJournal MCP is a Model Context Protocol server that connects AI tools to your team's derived work record. From Claude or any MCP-enabled assistant, you can ask what your team worked on, where the hours went and how something was done, with answers scoped to your role's permissions.

What is MCP?

MCP, the Model Context Protocol, is an open standard that lets AI assistants connect to external tools and data sources. An MCP server exposes a system's data in a form any compatible assistant can use, so the same connection works from Claude, MCP-enabled IDEs and agent tools alike. ScreenJournal MCP is ScreenJournal's implementation: it makes the team's derived work record available to the AI tools you already use, instead of asking you to switch windows to get an answer.

What can you do with ScreenJournal MCP?

Ask your own AI tools the questions you would ask ScreenJournal's built-in chat. Past activity is searchable through chat and MCP, so the record answers from wherever you work:

Planning in Claude: "what did the platform team ship last sprint, and what is still open?"
Preparing a client call: "where did the billable hours on this account go this month?"
Inside an AI coding tool: "how did we solve the webhook timeout in March?"
Running ops across time zones: "summarise yesterday for the Manila team."

Every query is permission-scoped by role. An employee can search their own history, a manager sees their team's derived record, and nobody sees more through MCP than they would see in the app.

How does ScreenJournal MCP work?

ScreenJournal MCP exposes the derived record, never the raw capture. It works in three steps.

The work is read and understood. ScreenJournal reads on-screen work as it happens; apps and sites on the privacy exclusion list are never captured. A frontier AI model derives a work timeline for each person, PII is removed during processing, and by default the raw screen data is deleted during processing.
Timelines accumulate into the chronicle. The work chronicle is the searchable history of the work itself: what was done, by whom, when and how.
The MCP server answers queries against that record. Connect an MCP client your company authorises, ask in plain English, and the answer comes back scoped to your role's permissions.

Can AI tools see screenshots through ScreenJournal MCP?

No, because there are no screenshots to see. ScreenJournal takes no screenshots, and by default stores no video either: the raw screen data is deleted during processing. Where a manager has opted in to keep video evidence, in what this site calls Evidence Mode, those recordings are stored for up to 3 months and replayed from the timeline in ScreenJournal itself; the MCP server has only derived insight to serve: timelines, summaries, numbers and answers. Keystrokes are not read in any mode. Apps and sites on the privacy exclusion list are never captured, and entries an employee has redacted are erased entirely, so they cannot surface through MCP or anywhere else.

What else can you do through ScreenJournal MCP?

Standups and summaries can be driven from a connected client. Ask for yesterday's standup and it is drafted from the timeline, in whichever model your team prefers, and the same connection answers questions about hours, attendance, app usage and productivity where the team already works. The connection is read-only: downloading Excel reports and the Timesheet stay in the dashboard.

Who is ScreenJournal MCP for?

Teams whose questions already start in an AI tool. Engineering leads who plan in Claude or an AI coding tool can pull real progress into the same window they are working in. Managers of distributed and offshore teams get answers ready when their day starts, in whichever assistant they use. Agencies can check billable reality without leaving the conversation they are in. And because MCP is an open standard, the record follows your tools as they change, not the other way round.

The built-in Ask AI does the same job inside ScreenJournal itself; MCP takes it wherever you work. For how an answer-first tool differs from footage-first alternatives, see ScreenJournal vs the alternatives.

Frequently asked questions

What is ScreenJournal?

ScreenJournal is an AI work visibility tool that reads on-screen work as it happens, turns it into a detailed timeline of what each person actually did, and then, by default, deletes the raw screen data. Timelines accumulate into a searchable chronicle of everyone's work history, and from them ScreenJournal generates timesheets and reports automatically and drafts standup summaries on request, answering questions about any of it in plain English.

Which AI tools can connect to ScreenJournal MCP?

Any client that supports the Model Context Protocol, including Claude and a growing set of MCP-enabled assistants, IDEs and agent tools. MCP is an open standard, so connecting does not depend on a tool-specific integration.

Who can see what through ScreenJournal MCP?

Exactly what they could see in the app, and no more. Queries run under the same role permissions as ScreenJournal itself, so employees can search their own history, managers see their team's derived record, and redacted entries are erased entirely.

Is ScreenJournal MCP available today?

Yes. Past activity is searchable through ScreenJournal's chat and MCP today, permission-scoped by role, and standup summaries can be drafted on request from any connected client. The connection is read-only: it answers questions about activity, hours, attendance, app usage and productivity, while building Excel reports and the Timesheet stay in the dashboard.