Give your AI agents the full context
Connect Claude, ChatGPT, Cursor, or a coding agent through the Screendesk MCP or the sdesk CLI. Agents read the recording, logs, and network requests, then investigate.
Trusted by support teams from startups to enterprises
Built for agents. Read-only unless you say otherwise.
Investigate across reports
Agents can find related reports, compare up to five, and summarize recurring errors, not just read one recording.
Writes only when you allow it
The CLI is read-only by default. Comments and report updates need --write and a confirmation.
Every report, readable by agents. The same context your team sees, through tools agents understand.
Spot recurring errors. Group the same failure across reports to see what’s breaking most often.
OAuth or a token. Connect agents with OAuth, or use a personal access token for scripts.
Redacted network data. Sensitive headers and values are redacted before agents see them.
Frames, screenshots, and transcripts. Agents can pull sampled video frames, screenshots, user events, and the transcript.
A developer API. Pull reports, recordings, transcripts, and analysis into your own tools.
Findings back on the report. With your confirmation, agents can comment on a report or update it.
Resolve
Turn the recording into action. Fix it, escalate it, or ask AI.
Resolve directly
Understand the issue and fix it without back-and-forth.
Escalate when needed
Send to Linear
Checkout fails after clicking Pay
I’m trying to pay for the Pro plan but the payment never goes through.
Share a complete report engineering can act on.
Investigate with Screendesk MCP
Send the full context to Claude, ChatGPT, Cursor, and more. Find the root cause. Suggest a fix.
Works better together. AI Agents is one part of the Screendesk platform.
Questions about ai agents
Which AI tools can connect?
Any MCP client, such as Claude, ChatGPT, or Cursor, through the Screendesk MCP server with OAuth. Coding agents and scripts can use the sdesk CLI.
What can agents read?
Report details, metadata, console logs, network requests, user events, transcripts, AI analysis, screenshots, and sampled video frames, plus folders and members.
Can agents change anything?
Only with explicit confirmation. Creating comments and updating reports are write tools; the CLI is read-only unless you pass --write and --yes.
How do I install the CLI?
Run curl -fsSL https://app.screendesk.io/cli/install | sh on macOS or Linux, then sign in with OAuth or a personal access token.
Is sensitive data protected?
Yes. Network data is redacted before it reaches agents, and access follows your workspace permissions.
