PendiumDocs

Work the recommendations

Get prioritized action items from a scan and mark them complete as you address them.

Turn a completed scan into a prioritized to-do list — content to create, sources to reach, and technical fixes — then keep the queue accurate by marking actions done as you go.

When to use

  • Reach for this after a scan completes, when you want the concrete next steps that will improve AI visibility rather than the raw scores.
  • Filter by status and priority to focus on the most impactful, still-open work.
  • This job depends on a scan existing — run one first via Run a visibility scan, and re-scan afterward to measure impact.

Typical sequence

  1. get_recommendations — pull actions, typically filtered status: "pending" and priority: "high".
  2. Gather context for the work with get_brand_voice, get_factsheet, and query_knowledge.
  3. complete_action — mark each action done once you've addressed it.
  4. Re-run a scan with scan_visibility to measure the impact.

Tools

get_recommendations

Get prioritized action items to improve AI visibility. Actions are generated from scan recommendations and cover content to create, sources to reach out to, and technical improvements.

Authentication: Required

ParameterTypeRequiredDescription
syntheticIdintegerYesThe agent ID to get actions for.
status"pending" | "in_progress" | "completed" | "dismissed"NoFilter by action status. Returns all statuses if omitted.
priority"high" | "medium" | "low"NoFilter by priority level.

Returns actionCount, pendingCount, and an actions array. Each action carries id, actionType, category, title, description, priority, status, targetUrl, relatedQueries, createdAt, and completedAt.

Action types:

TypeDescription
create_contentWrite a blog post, article, or page targeting specific queries.
update_contentImprove existing content to better answer AI queries.
outreachReach out to sources that AI platforms cite.
technicalTechnical SEO or structured data improvements.

complete_action

Mark a visibility action as completed after you've addressed it — created the content, made the outreach, or implemented the technical fix. Keeps the action queue accurate.

Authentication: Required. An account in read-only mode can't do this (it's a write) — the call returns a read-only error.

ParameterTypeRequiredDescription
syntheticIdintegerYesThe agent ID the action belongs to.
actionIdstringYesThe action ID to mark as completed. Get action IDs from get_recommendations.

Returns success: true with the updated action (id, title, status: "completed", completedAt). If the action ID doesn't exist or doesn't belong to the specified agent, the tool errors with an "Action not found" message.

On this page