PageLens AI exposes MCP tooling so AI assistants can inspect reports and help you turn findings into implementation work.
MCP is different from scan automation. Use MCP when an assistant needs to read or comment on report data. Use API keys, CLI scans, GitHub Actions or deploy hooks when you want a trusted workflow to create scans. See Developer automation for the scan automation overview.
What MCP is for
Use MCP when you want an assistant to:
- read a scan report
- inspect findings and evidence
- summarize what changed between scans
- plan fixes from report data
- flag a finding that looks wrong
- record that a real finding is an intentional decision
Finding feedback
Agents can submit finding-level feedback for false positives, wrong severity, wrong category, or findings that are not actionable. These flags go to the PageLens AI admin queue for manual review.
This is separate from support tickets. Use finding feedback when the issue is about a specific finding. Use support when you need help with billing, scan setup, authenticated routes, or a product question.
Accepted decisions
Sometimes a finding is correct, but you have made a deliberate product or architectural choice. For example, you may accept a Content Security Policy tradeoff because a payment or component library needs inline styles.
In that case, an MCP-connected assistant can record an accepted decision against the finding. PageLens AI keeps the finding visible, keeps the score unchanged, and shows the decision as context on the report. Future scans can also match the same decision when the domain, page, viewport, rule, and finding evidence line up.
Accepted decisions are customer-owned notes, not false-positive triage. They are useful for explaining intentional tradeoffs to your team and for helping an assistant avoid re-litigating the same decision every time it reviews the report.
If the decision changes, the assistant can clear the decision so it no longer appears on current or future reports. PageLens AI keeps the audit trail.
Permissions
Only connect tools and agents you trust. OAuth grants can be revoked, and PageLens AI records enough audit trail to understand which client submitted feedback or recorded a decision. These tools cannot hide findings, change scores, delete scans, or spend money.
Best practice
Ask the assistant to cite finding IDs, page URLs, and evidence when it questions a result or records a decision. That gives the PageLens AI team enough context to review feedback, and gives future agents enough context to understand why a tradeoff was accepted.