GEO

Traces

Automatic inference calls to track LLM citation and mention rates.

The call
$ letterstory probe add \
  --prompt "best RTL verification tools"
list_probes · this week
MCPCLIREST
How it runs
  1. Ask

    The same prompt to every major answer engine, daily.

  2. Parse

    Citations, rank and competing brands pulled from each answer.

  3. Trend

    One series per prompt, so a change is a line, not a hunch.

Shape

Takes

promptscopeengines
→

Returns

citationsrankcompetitorstrend
list_probes · this week
cited in
2 of 3 engines
rank
4 → 2
taken by
1 competitor
prompts
24 tracked
example output
Watch it from an agent
  • get_competitor_feed
  • get_competitor_topics
  • get_phantom_geo_signals
  • get_phantom_seo_article
  • get_phantom_seo_overview
  • webhookprobe_run.completed
Tools
35 in the manifest · one schema on every surface
add_competitoradd_probeadd_ranking_targetbackfill_seo_historycontrol_probeget_competitor_feedget_competitor_topicsget_phantom_geo_signalsget_phantom_seo_articleget_phantom_seo_overviewget_phantom_seo_signalsget_post_performanceget_recent_winsget_seo_cluster_performanceget_seo_performance_trendget_seo_query_performanceget_site_performanceguess_geo_competitorsinfer_competitor_keywordslist_article_performancelist_competitorslist_geo_opportunitieslist_phantom_seo_articleslist_phantom_seo_siteslist_probeslist_ranking_targetslist_seo_opportunitieslist_top_postsremove_competitorremove_seo_opportunitiesscrape_competitorsuggest_ranking_targetssync_geo_opportunitiesupdate_competitorupdate_ranking_target
Pairs with
Traces → Writing kernels

Adversarial harnesses for producing high-quality writing that ranks on search.

Traces → Seers

Listeners that trigger content from evolving sources of truth.

← PreviousWriting kernelsNext →Proofs