Skip to main content
Internal links scans your website and saves evidence for you and your agent to review. It does not publish changes. Background scans stop at the page limit authorized for your workspace and saved for the run. 300 pages is the default; higher limits require separate authorization. Check Scan details for coverage and the applicable allowance. The scan follows pages on the exact hostname. Where enabled, limited browser rendering can supplement captured HTML; JavaScript links, subdomains and undiscovered pages may still be missing. Blocked or incomplete checks stay unknown. Starting one scan does not create a rescan schedule or enroll backlink monitoring.

Start a scan and return to its saved progress

  1. Choose your website, open Internal links and select Scan website. Review any allowance or credit confirmation before starting.
  2. Watch the saved progress move from Queued to Scanning. You can leave the page while it runs.
  3. Open Scan details and choose a Saved scan. Use Refresh saved evidence in the page menu to read progress without starting another scan.
  4. Open Health. Use View to switch between Suggestions and Findings, then choose Review on an item.
The progress summary separates checked, unknown and waiting pages. Finished within scope means the admitted work finished, not that every page on the website was found. Open Scan details and read Coverage and limits. Check the saved run’s expiry date and download evidence you need to keep. Background scans retain up to 20 runs per website for up to 30 days; using an older saved page list can shorten that expiry. Manual page lists have a separate 20-record limit, which can include older scans.

Read findings as questions to investigate

In Health, choose Findings from View. Select Review on a finding, then Inspect connections. Start with broken destinations or important pages that appear poorly connected. Check coverage before treating an orphan candidate as a whole-site fact.
No findings does not establish that the whole website is healthy. Check the inventory coverage and unknown pages first.

Inspect the connection behind a finding

In Site map, enter a Page to inspect and select Inspect page. Open a connection row for its link text, surrounding passage and observation time. Use Open source page to check the current page before editing. A destination outside the scan inventory can appear in a connection, but its check status and outgoing links remain unknown. The graph shows up to six incoming and six outgoing pages from the loaded results. Use Next evidence for more rows. Show contextual links only initially filters to links found in HTML main-content landmarks; check the page yourself to judge relevance and visibility.

Give your agent evidence and a specific task

Choose Work through suggestions to open an agent handoff. For a specific page, use Prepare an agent task for this page in Site map. The page menu also offers Copy task for your agent; if copying fails, use the displayed Agent task text. Ask the agent to read the saved evidence, inspect current content and propose a link in an exact passage. Review and publish through your website’s normal editing process. Copying the task does not publish or schedule anything. Download scan evidence saves pages, connections and findings across all result pages. If the scan changes during export, refresh and retry. Sync and export covers other saved records.

Pause, resume or recover a scan

Pause scan stops further work; a request in progress may finish. Resume scan continues with current access. Cancel scan ends the run and keeps saved results. Start a new scan after cancellation or completion. If a run stops for a daily page allowance, follow its displayed reset and choose Resume scan. If the run stops for credits or billing, follow that status rather than assuming a midnight reset. Reading progress does not resume it. For page allowances, pages count when admitted, even if a check fails. Each run stays within its saved page limit. If Rescan this website is available, choose Every week or Every month for repeat scans. Review any allowance or credit confirmation before enabling them. Choose Only when I start a scan to turn that schedule off. Cancelling a run does not turn off an existing schedule.
Use monitoring for earned links you want checked over time.From Needs you, open a stopped scan to read its Health view for the correct website. Opening it does not resume work. Discovery respects robots restrictions and skips non-page files, outside addresses and indefinitely repeating URL patterns.

Choose particular pages instead

Use Choose pages manually from the page menu when you want to check a supplied list. The default allowance for manual lists is 300 pages per workspace per UTC day, shared with other page-allowance scans. Follow the allowance or credit status shown for your workspace. Save page list reserves the listed pages against that allowance without fetching them; Check these pages starts checks. Unlike a background scan, closing this manual workflow stops the app from starting further batches.
Open Pages to select or supply the list.Select Discover my pages to preview URLs from robots.txt and up to three sitemap documents. Discovery looks for up to 100 page URLs for up to 20 seconds, uses no page-check allowance, and starts no monitoring. A workspace can start three previews per minute; a preview lasts 15 minutes.Review Choose pages to check, clear unwanted URLs and mark any Priority page. Open Sitemap sources and coverage to inspect the source details. Check selected … pages saves the selected list and starts its checks. If confirmation was interrupted, use Retry the same selection. An already-saved selection can be recovered after preview expiry; if nothing was saved, discover pages again.To supply URLs yourself, expand Paste page URLs instead. Paste up to 100 distinct page URLs with the website’s exact hostname. Add priority pages or up to 10 starting pages from that same list. Mark the list complete only if it is complete; AgentLinkOps records your declaration without verifying it. Save page list saves the inventory without fetching pages and counts those pages toward the shared daily allowance.Pause checks stops after the current batch; Continue page checks processes the remaining pages. Return to the saved list to continue. Follow any displayed wait time; checks can also pause after ten minutes. To recheck processed pages, save a new list.

Run a background scan through commands

Start with projectId and a caller-generated requestId. After an uncertain start response, reuse the same requestId for the same website. Keep the returned scanId for reads and controls. Reading or listing saved results does not start, resume or schedule work.For a control request, pass revision from the latest scan.controlRevision, along with projectId, scanId and action. If the controls changed, refresh before deciding whether to retry. Follow query cursors with the same scan and filters. If the scan changes during pagination, refresh and restart the read.

Commands for a selected page list

For a sitemap selection, pass its previewId and selected URLs to create_internal_link_snapshot, with inventoryComplete: false. Give preview and save operations separate requestId values; reuse the same operation’s value and inputs after an uncertain response. Saving does not check pages. Call process_internal_link_snapshot to process them, preferably with maxPages: 1 for interactive clients, and respect returned retry timing.Reads require watches:read for the website. Starting, controlling, discovering, saving or processing also requires watches:write and a role allowed to manage links. The app’s download action also requires exports:create. Current workspace membership and website access apply to requests and continuing background work.