Example prompts and workflows for Astra MCP

Last updated: November 2, 2026

This article lists example prompts and multi-step workflows for Astra MCP, plus tips for getting accurate answers.

Replace the text in square brackets with a target name or a vulnerability title from the Astra account. Prompts work in any connected AI tool.

Monitor security posture

  • "Which targets have not been scanned in the last 30 days?" Returns the targets that have not been scanned recently, with their latest scan details.

  • "Give me a snapshot of the account's security posture." Returns an overview of the targets, recent scans and open vulnerabilities.

  • "What is the status of the latest scan on [target]?" Returns the scan's current status and statistics.

  • "When is the next scheduled scan?" Returns the recurring schedules and when each one runs next.

Triage vulnerabilities

  • "Show the open critical and high vulnerabilities, grouped by target." Returns each vulnerability with its title, severity, status and target.

  • "Which vulnerability should be fixed first, and why?" Returns the open vulnerabilities ranked by severity, to help decide where to start.

  • "Explain [vulnerability] in plain language and describe how to fix it." Returns a summary built from the vulnerability's details, findings and comments, with suggested fix steps. If the AI tool has access to the project's code, ask for a concrete code change.

Collaborate with the team

  • "Show the comments on [vulnerability]." Returns the existing discussion.

  • "Add a comment to [vulnerability]: 'Fix released in version 2.4, ready for review.'" Adds the comment to the vulnerability in Astra.

Verify fixes

  • "Mark [vulnerability] as fixed and request a rescan." Moves the vulnerability to the next step in its workflow, checks that a rescan is possible, and requests it. If a rescan is not allowed, the assistant explains why.

  • "Show the results of the rescan for [vulnerability]." Returns the rescan status and its outcome.

Start and schedule scans

  • "Start a scan on [target]." The assistant asks which scan type to use (Full, Emerging or Lightning) and, for a Full scan, the coverage to use. It does not choose these on its own. It then returns the scan ID, target and status.

  • "Set up a recurring scan for [target] every Monday at 9 AM." The assistant asks for the scan type, then creates the schedule and returns the next run time.

  • "Cancel the running scan on [target]." Cancels the scan if it can still be cancelled.

  • "Delete the schedule for [target]." The assistant shows the schedule and asks for confirmation before it deletes anything.

Generate and share reports

  • "Generate a pentest report that can be shared with stakeholders." The assistant asks which report type is needed (for example Full Report or Management Report), which scans to include, and whether the report should be emailed or only generated. It returns the report details and a link to open it in Astra.

  • "Find the most recent reports." Returns the latest reports with their type and status, and a link to each one in Astra.

Note

Report files are not downloaded through Astra MCP. The assistant shares a link that opens the report in the Reports tab in Astra, where it can be downloaded.

Multi-step workflows

Each step builds on the previous answer, so send the prompts one at a time in the same chat.

Fix a vulnerability from start to finish

  1. "Show the open critical and high vulnerabilities, grouped by target."

  2. "Explain the first vulnerability and describe how to fix it."

  3. "Add a comment to it: 'Fix is in progress. Owner: backend team.'"

  4. "The fix is deployed. Mark it as fixed and request a rescan."

Tip

In an editor such as Cursor, VS Code or Claude Code, ask the assistant to fix a vulnerability, let it read the remediation brief, comments and findings, and review the code change. Then ask it to add a comment and Submit for review. A rescan can then confirm the fix. AI tools that support MCP prompts can start this flow with the astra_fix_vulnerability prompt.

Prepare a security update for stakeholders

  1. "What did the last scan on [target] find, and how many vulnerabilities are still open?"

  2. "Summarize this for a non-technical audience in five bullet points. Do not overstate progress."

  3. "Generate a management report for that scan and share the link."

Review the current status of a target

  1. "What is the status of the latest scan on [target]?"

  2. "Show the vulnerabilities that are still open, most severe first."

  3. "Do any of them have new comments?"

Tips for better results

  • Be specific. Use the target name or vulnerability title, and state which one is meant when there are several.

  • Add a time frame or an order. "In the last 30 days" or "most severe first" produces a more focused answer than "show me everything".

  • Specify the format. A short summary, a table or a list of next steps all work.

  • Request a preview before changes. For any action that adds, updates or deletes, ask the assistant to show the change and wait for approval before it proceeds.

  • Ask the assistant not to guess. Add an instruction such as "if you are not sure, say so instead of guessing".

  • Work in steps. Build on each answer instead of requesting everything in one prompt.

Important

AI output can be wrong. Review the assistant's answer before acting on it, especially before starting a scan, changing a vulnerability status or deleting a schedule.

For anything not covered here, contact help@getastra.com.