EdgeComet glossary: terms used in the dashboard and docs

This glossary defines the terms EdgeComet uses in its dashboard and docs, grouped by product area. Each term links to the page that explains it in full.

General SEO terms, such as crawl budget, keyword cannibalization or server-side rendering, are explained on their own pages: Crawl budget and crawl waste explained and How Googlebot and AI bots handle JavaScript.

Setup and serving

Terms for connecting a website and serving its bots.

Term	Meaning
Integration health	EdgeComet's check, about every 15 minutes after a site goes live, of whether verified bots keep arriving, shown as Active, No recent traffic or No traffic 7d+.
Live	The state a site reaches once EdgeComet has served at least one response to a verified bot, and keeps after that. Not the same as a Live Edge SEO rule.
Render host	The EdgeComet address your routing rule sends bot requests to, render-us.edgecomet.com in the US or render-de.edgecomet.com in Germany, shown in every setup guide.
Render key	Your site's private key, sent in the X-Render-Key header, that identifies requests coming from EdgeComet and that your firewall allow rule trusts.
Routing rule	The rule at your CDN or proxy that sends page requests from crawlers to EdgeComet and every other request, including failures, to your origin.
Serving mode	The setting that decides how EdgeComet builds the pages it serves to bots: Cache-only stores your server's HTML, and Cache + Render runs the page's JavaScript in headless Chrome first.

Cache

Terms for how EdgeComet stores pages for bots and keeps them fresh.

Term	Meaning
Bypass cache	The cache of your server's own HTML, stored for dimensions that bypass rendering, on for Cache-only websites and off on a new Cache + Render website.
Cache TTL	How long a cached copy is served before EdgeComet builds a new one, 7 days on a new website and from 1 minute to 30 days.
Concurrent re-caches	How many pages EdgeComet refreshes at the same time for your website, 5 by default and 1 to 50, which limits refreshes only and not live bot requests.
Flush All	The action on Cached URLs that removes every cached copy of the website on every dimension; it cannot be undone and does not clear the pre-cache queue.
Invalidate	Deleting a cached copy at once, shown as Remove in the dashboard, which makes the next bot request wait for a fresh render or fetch.
Pre-cache	Rendering or fetching your pages on a schedule and storing them, which gives Googlebot and AI bots a fresh cached copy instead of a wait for a render or your server.
Pre-cache rule	A per-website rule that renders or fetches a set of pages on a schedule, such as Smart Re-cache or Sitemap Re-cache, and takes effect on Save with no deploy.
Pre-cached HTML	The choice on every pre-cache rule of what each copy stores: Original HTML as the website normally serves it, Rendered HTML from headless Chrome, or a choice per copy.
Priority pages	The pages Smart Re-cache sends first on every run at high priority, changed or not: the homepage, top Search Console pages, main homepage links and URLs you add.
Re-check interval	The per-page interval Smart Re-cache learns from the page change log, which grows while the page stays unchanged, resets on a change and never passes the Cache TTL.
Recache	Building a new cached copy of a page while EdgeComet keeps serving the current one until the new copy is stored.
Render cache	The cache of pages EdgeComet rendered in headless Chrome, which is always on.
Smart Re-cache	The main pre-cache rule, which refreshes each page bots request as often as that page changes, and your homepage and top pages on every run.
Stale copy	An expired cached copy EdgeComet keeps as a fallback and serves only when a fresh render or fetch fails, for up to the Stale TTL.
Stale TTL	How long EdgeComet keeps an expired copy as a fallback before deleting it, with Serve stale content if origin is down on; 24 hours by default, 1 hour to 7 days.
Warm on last request	A URL bots requested that was served from the cache the last time a bot asked, as counted per bot group in Bot coverage.

Rendering

Terms for how EdgeComet renders JavaScript pages and which settings control it.

Term	Meaning
Block dimension	A dimension with the Block action, which answers matching bots with 403 Forbidden for every URL, before any URL rule, cache lookup or render.
Blocked resources	The render settings Blocked AJAX URL Patterns and Blocked Resource Types, which block matching requests during a render; Image, Media and Font are blocked by default.
Bot alias	A named set of user agent patterns that EdgeComet keeps up to date, such as Anthropic Bot, which you pick for a dimension instead of writing a regex.
Bypass dimension	The one dimension every website has that sends your server's HTML without rendering, catches every user agent no other dimension matches, and cannot be deleted.
Deployment history	The list of every configuration deploy under Configuration > System Settings > Deployment History, with its date, author and changes, from which you restore an earlier configuration as a draft.
Dimension	A device profile that decides how EdgeComet builds the copy a bot gets: its viewport, the user agent Chrome renders with, and which bots it serves.
Draft configuration	Saved changes to System Settings or Caching > Settings that bots do not see until you deploy them; a deploy applies the whole draft to live traffic at once.
Failed render	A render that ended in an error, other than a soft timeout or a 4xx from your server, or that returned a 5xx.
Fresh render	A render that returned status 200 with no error, counted on the Fresh Renders card of Cache > Renders.
Hard timeout	A render that exceeded the rendering service's maximum time, with no HTML captured; bots got the last cached copy or your server's HTML instead.
Render error type	The one error type, such as Soft Timeout, Origin 5xx or Pool Unavailable, that each failed render or pre-cache attempt carries, shown with its severity on the Issues tab.
Render Timeout	The render setting for the longest a render may take, 15 seconds by default; when it runs out before the wait event, EdgeComet serves what the page shows, a soft timeout.
Rule evaluation order	The fixed order that picks one URL rule when several match: user agent rules, then Exact, Wildcard and Regex patterns, query conditions, deeper paths, then the order you added them.
Soft timeout	A render whose wait event did not fire before the Render Timeout; EdgeComet still captured and served the page as it was, which may be incomplete.
Strip Scripts	The render setting, on by default, that removes executable script tags from the rendered HTML and keeps JSON-LD structured data.
URL rule	A rule under Configuration > System Settings > URL Rules that pairs URL patterns with an action, such as Render, Bypass, Block or a status code, and optional settings.
Wait For Event	The render setting that decides when Chrome captures the page, such as DOMContentLoaded, load or networkIdle, the default.

Bots and Log Analyzer

Terms for the bots EdgeComet recognizes and the bot data in Log Analyzer.

Term	Meaning
AI attention map	A card in Log Analyzer > AI Bots that splits one AI provider's requests, from ChatGPT, Claude or Perplexity, by purpose (Training, Search, User) and by segment of your site.
AI search bot	A bot that crawls pages for an AI assistant's search index, as a search engine crawler does for a search engine, such as OAI-SearchBot and PerplexityBot.
AI user bot	A bot that fetches a page because a person asked an AI assistant something, not on a crawl schedule, such as ChatGPT-User and Claude-User.
Bad bot	A bot whose request failed IP verification, an unrecognized crawler, or a known SEO tool: traffic you may not want, not proof of harm.
Bot category	The kind of bot a request came from, one of six categories such as Search Bots, AI Bots and SEO Tools that EdgeComet assigns to every recognized bot.
Bot family	One of seven non-overlapping groups of bots, such as Googlebot Search, ChatGPT or Fake Bots, that the Log Analyzer Bot Activity chart draws as one line.
Bot variants	The crawlers of one search engine, such as Googlebot Mobile, Googlebot Image or Bing Preview, listed one per row in the Bot Variants table on the Crawl Budget tab.
Catch-all page	The one page your site serves with a 200 for any unknown path, such as a single-page app shell or a bot challenge page; probes that get it count as Same Page.
Chat share	One request from iMessage or WhatsApp, which Log Analyzer > Social Bots counts as one paste of the link and shows as iMessage & WhatsApp.
Crawl budget allocation	EdgeComet's split of the bot requests it served into Wasted, Questionable and Valuable, where Valuable is every request no rule matches.
Crawl efficiency	How a bot family's requests split into valuable, questionable and wasted crawls, shown on the Efficiency tab of Log Analyzer > Googlebot, Bing and AI Bots. Not the Crawl Efficiency card on the Log Analyzer Overview, which shows the share of 2xx responses.
Declining crawl	Pages the bot requested last period that got at least 50% fewer requests this period, including pages it stopped requesting.
Delivery performance	How much of Googlebot's or Bingbot's crawl EdgeComet answered from its cache, and how fast cache and render requests were served, on the Delivery Performance card.
Event type	The field of each Bot log record that says how EdgeComet handled the request: Cache Hit, Render, Bypass, Bypass Cache, Precache or Error.
Exposed file	A probed path that answered 200 with content in the period, other than your site's catch-all page or a login page.
Fake bot	A request that names a bot EdgeComet can verify, such as Googlebot or GPTBot, from an IP address outside every range that bot's operator publishes.
Hidden request	A social bot request left out of every number on the Social Bots page: a spoofed app, a half-typed link from WhatsApp or Telegram, or a Meta or Apple platform check.
IP validation	EdgeComet's check of each bot request's IP address against the ranges the claimed bot's operator publishes, stored as the IP Validation field: Verified, Fake, Unverifiable or Error.
New URLs found	Pages the bot requested this period but not in the previous period of the same length, counted on the Efficiency tab.
Overcrawled	Pages above this period's 99th percentile of requests per page, roughly the busiest 1% of the pages the bot requested.
Page size band	A range of HTML sizes, from 0-100 KB to >5 MB, that the Page Size tab uses to group 200 responses; page size counts the HTML document only.
Probe outcome	The result of one probe request, from its status code and content: Exposed, Same Page, Server Error, Login Reachable, Redirected, Blocked or Not Found.
Probe request	A request for a path on EdgeComet's list of paths scanners use to find files and endpoints that must never be public, such as /.env or /.git/config, or a request whose query string carries a scanner payload. Not the same as a fan-out probe.
Render weight	Everything the browser downloaded while EdgeComet rendered a page: the HTML plus scripts, stylesheets, images and fonts.
Request-path capture	Recording each request inside the service that answers it, at the moment it answers, which is how EdgeComet collects bot data without server log files.
Risk tier	One of three groups that sort bad bot requests by why they count: High Risk (Fake), Medium Risk (Unknown) and Low Risk (Known).
Serve time	The time EdgeComet takes to answer a bot request, from the start of handling to the ready response, including the cache lookup and any render on request.
Shared page	A page any social bot requested in the period, counted once across its query-string variants. Each Needs Attention share on the Social Bots page is a share of these pages.
Social bot	The fetcher a chat or social app, such as iMessage, WhatsApp, Facebook or LinkedIn, sends to build the preview card under a shared link.
Status class	The first digit of an HTTP status code, which Log Analyzer uses to group codes: 2xx success, 3xx redirect, 4xx client error and 5xx server error.
Training bot	A bot that collects pages that may be used to train AI models, such as GPTBot and ClaudeBot.
Unique URLs	The count of distinct pages after URL normalization, the step EdgeComet uses for cache keys, which removes tracking parameters and sorts the rest.
URL inventory	A card on the Efficiency tab that compares the pages the bot requested this period with the previous period, as new, lost and returning URLs by kind.
Verified bot	A bot request whose IP address falls inside the ranges the bot's operator publishes, recorded as Verified in the IP Validation field.
Waste condition	One of six fixed checks, such as 404 Not Found or Blocked but Crawled, listed as the Issue Type rows of the Crawl Waste Breakdown table.
Wasted renders	The renders EdgeComet ran in a browser to answer bad bot requests; requests answered from the cache need no render.

Evergreen Crawl

Terms for Evergreen Crawl, the site audit built from what bots received.

Term	Meaning
Canonical disposition	The Canonicals report table that puts each 200 page in one bucket, such as Self-canonical or Cross, broken target, from its declared canonical and the target's health.
Comparison crawl	The earlier crawl EdgeComet picks for each report's change figures, with the same bots and window length, starting no later than the crawl you read.
Crawl issue severity	The fixed level of each Evergreen Crawl check: P0 active loss or config conflict, P1 newly broken with demand, P2 strategic, P3 hygiene.
Current crawl	The crawl EdgeComet rebuilds every day from the last 30 days of bot visits and pre-cache fetches.
Custom crawl	A crawl you start with + New crawl for the bots and period you choose, built in minutes from bot visits EdgeComet already recorded.
Declaration profile	One distinct set of languages that pages declare in hreflang, where one profile across the site means every page declares the same set.
Demand	The traffic a page carries in the crawl's window: its Google impressions from Search Console and its fetches by AI assistants answering users, shown side by side and never added. Not Google's crawl demand.
Duplication cluster	A group of pages with a duplicated tag that one fix clears, either a failing template or one repeated string with no template behind it.
Evergreen Crawl	EdgeComet's site audit built from the pages search engines and AI bots actually received from your site, without a crawler of its own.
Index leak	An indexable page with no Google impression and no AI assistant fetch in the crawl's window.
Indexability bucket	One of the Indexability table's classes, such as Indexable, Noindex, Canonical to other or Redirect 3xx, decided by status code, then noindex, then the canonical.
Issue category	One of the 11 groups Evergreen Crawl checks fall into, such as Canonicals, Hreflang or Catalog Health, each with its own group in the All issues lens.
Link-only stub	A URL that crawled pages link to but no bot fetched in the window, kept in the crawl with its inlinks but without its content.
Missing expected markup	The Structured Data card counting product pages without a JSON-LD Product and articles without an Article, including pages where it is only nested inside another node.
Monthly crawl	A crawl EdgeComet makes in the first days of each month over the 30 days before, named for the previous month, such as September 2026.
Starved winner	A page with only 1 or 2 content links that still earns Google impressions or AI fetches.
Structured data severity	The level of each validation finding: Critical blocks a Google rich result, Warning marks problems such as an invalid value or unknown property, and Opportunity a missing recommended property.
Surfaced page	An indexable page with at least one Google impression or one AI assistant fetch in the crawl's window, evidence that something showed it, not proof that Google indexed it.
Templated duplicate	A title, meta description or H1 shared with another page and stamped by a template whose variable part fails to differentiate, fixed by one edit per template.
Validation finding	One property and one problem on one structured data type, such as Article › dateModified: Recommended property missing, counting every page that carries it. Not the same as an Action Board finding.

Search

Terms for the Search Console reports and fan-out queries.

Term	Meaning
Coverage check	A check on the Page Analysis Fan-out tab that grades whether a page, as EdgeComet serves it to bots, answers each question AI assistants searched for.
Diluted clicks	A cannibalized keyword's total clicks minus the clicks of its leading URL, read as the most that combining the pages could recover, not as measured loss.
Fan-out keyword	A Search Console keyword that EdgeComet detects as a search an AI assistant ran on a person's behalf.
Fan-out question	The question a person most likely asked, written by AI for a group of fan-out keywords that ask the same thing in different words.
Fan-out threshold	The card that shows what EdgeComet measured on your site to detect fan-out keywords, tuned from your own Search Console data with nothing to set.
Fan-out topic	A subject discovered from your site's own fan-out questions; each question belongs to at most one topic.
Keyword movement	The Search Overview card that splits the change in clicks against the previous period into new, improved, declined and lost keywords, grouped by click change, not position change.
Keyword opportunity	A keyword and page pair in one of four buckets, Striking distance, Below-curve CTR, Zero-click high-impression or Cannibalized keywords, each a reason it earns fewer clicks than it could.
Probe	A fan-out keyword that an automated AI tool or a scraper produced, not a person's question, shown in its own lane under Who is asking. Not the same as a probe request from a security scanner.
Subfolder property	A Search Console URL-prefix property for one section of your site, such as https://example.com/blog/, merged with the root property with every page counted once. Each root property includes 10 at no extra cost.
Trash keyword	A rank-tracker artifact shaped like "12 : some keyword", not a real search query, which the Trash keywords filter keeps or removes.

Changes

Terms for the page change log and site releases.

Term	Meaning
Change log modes	The Changes and Pages switch above the Change log table: Changes shows one row per change with old and new values, and Pages one row per changed page with its recrawl state.
Change pattern	One kind of change in words, such as "Status 200 → 404" or "Title rewritten"; the Changes table has one row per change pattern on one page type.
Page change log	The log EdgeComet builds by comparing each new copy of a page with the one before: one entry per change, with the page, field, old and new values, and time.
Page type (Changes)	A segment that sorts your URLs, such as product pages, listings or articles, which the Changes table and Recrawl use to group pages; others go under Other pages.
Priority (Changes table)	The default order of the Changes table, which sorts change patterns into five bands, losses first and recoveries last, with the most affected pages first in each band.
Recrawl state	The state each counted changed page gets per crawler, from that crawler's last verified fetch: Fetched after change, Unclear, Change unseen or Page unseen.
Release class	The Loss, Edit or Gain pill beside each site release.
Release status	Where each page type of a site release stands: Rolling out, Complete, Reverted, Loss reverted, Partly reverted, Flipping back and forth or Expected.
Site release	A deploy that made the same change on many pages of one page type, which EdgeComet finds on its own and follows while it rolls out, such as removing every product H1.
Tracked fields	The 11 fields Changes logs on every page, from status code and canonical to title, hreflang and the page body, plus each single-value custom extraction field you track.
Unstable pages	Pages where one field changed 3 or more times in the period and ended on the value it started with.

Alerts

Terms for alert rules, incidents and the weekly email.

Term	Meaning
AI reading	EdgeComet's AI review of a new incident, which writes its title, summary and fix steps and gives a verdict: Likely cause, Looks intended, Mostly noise or On EdgeComet's side.
Alert rule	A rule, which the dashboard also calls an alert, that says what to count, on which pages and at what threshold, and opens an incident when the condition holds.
Alert template	A ready-made alert, with what it watches, a default threshold, a severity and how often it checks, that you add a rule from.
Alerts	The EdgeComet feature that tells you when something breaks for search engines and AI bots, reading data EdgeComet already records, with no monitoring crawler to schedule.
Cooldown	The time after a rule fires, one hour by default, in which repeat matches are held or join the open incident instead of sending a new email.
Impressions at risk	The Search Console impressions an incident's affected pages earned in the 28 days before it fired, a measure of search exposure, not of lost traffic or revenue.
Incident	What opens on the Alerts page when an alert rule's condition holds, listing the affected pages, and goes to your destinations at the severities each one takes.
Index status	EdgeComet's own reading of whether the page it received can be indexed, from its status code, robots directives and canonical, not a report of what Google has indexed.
Pages changed	The alert that fires when a fact search engines read, such as a status code, indexability, canonical, title or extracted value, changes on many pages at once.
Rule state	The one state each alert rule is in, from its switch and its last check: Firing, Watching, Warming, Skipped, Broken or Paused.
Severity	The level every alert rule has, shown as a P badge: P0 Critical and P1 Warning are notified, and P2 Info appears in the product only. Crawl issues and feed checks use P levels with their own meanings: see Crawl issue severity and Feed check severity.
Site analysis	The AI agent that drafts the alerts your site should have from 30 days of bot traffic, the latest crawl, your segments and Search Console, with thresholds set from your numbers.
Weekly performance email	A weekly scorecard email of each site's Google clicks and impressions, AI answer fetches and Googlebot crawl against the week before, sent to the account owner every Wednesday.

Action Board

Terms for the Action Board, the work list the AI SEO agent keeps.

Term	Meaning
Action Board	The SEO work list the AI SEO agent keeps for your website: it finds what to fix, ranks it by the search and AI traffic at stake, and prepares the fix.
Action Board analysis	The run, about every 14 days, in which the AI SEO agent reads your site's crawl, Search Console and bot data and writes the Action Board.
Developer brief	One document that hands an SEO finding to your engineers, with the problem, the evidence, the recommended fix and how to confirm it, written when you choose Create dev task.
Document	A file the AI SEO agent writes for you, such as a content brief, developer brief, article edits, research write-up or export, listed under Action Board > Outputs > Documents.
Finding	One job on the Action Board: every page on it gets the same instruction, and it can have several steps, each with its own pages and fix.
Fix route	One of the ways a finding offers to fix it: serve new values at the edge, hand a brief to your developers, brief your writers, or track the work yourself.
Generation run	The AI job that writes a fix's drafts, such as new page values, a developer brief or content briefs, none of which reach your site until a person deploys them.
Lane	A group of Action Board findings by who fixes them: Fix at the edge, For your developers, For your content team, or Watch.
Page brief	A short content brief for one existing page, shown as Content brief in the dashboard: what the page does today, where it falls short and what to improve.
Proposed finding	A finding that Ask or a skill run recorded in a conversation and that is not on the board yet, shown as a Proposed finding card with Add to board.
Task	The fix you committed to on an Action Board finding: values deployed at the edge, a developer brief, content briefs, or a follow-up you track by hand.
Verification	How EdgeComet confirms a fix worked: after you deploy or ship it, the next recount checks whether any of its pages are still in the state the finding fixes.

Ask

Terms for Ask, the AI SEO agent you chat with.

Term	Meaning
Ask	The EdgeComet AI SEO agent, which answers questions about your website in plain words from your own data: bot fetches, Search Console performance and the Evergreen Crawl.
Ask panel	A chat beside any report, opened with Ask in the report header or Cmd+. (Ctrl+.), whose answers read that report's numbers for the period and segment on screen.
Conversation	One Ask thread of questions and answers about a website, kept together with the tables, charts and documents it produced.
Custom report	A report you build by describing it in Ask, made of 1 to 12 blocks that read live data for the period and segment on screen.
Custom skill	An SEO workflow you write once in plain Markdown and run on demand from Ask, which the AI SEO agent follows with the same data and tools as the built-in skills.
Focus	A limit on Ask questions to pages you choose, up to 20 pages, 200 URLs or a path prefix, that stays on for your next questions until you remove it.
Guidelines	Standing instructions about your business that the AI SEO agent applies on every run: Action Board analyses, Ask answers, skill runs, rewrites and briefs.
Live table	An Ask table that is the result of a query the engine runs, which re-runs when you change the period, segment or filter and sorts on the server.
Report block	One card of a custom report, a stat, line, bar, pie or table, that holds its own query; a report holds 1 to 12 blocks.
Run-inputs block	A fenced block at the top of a custom skill's playbook that lists the questions the Run dialog asks before the skill starts, and can pin where the skill runs.
Skill	A saved SEO workflow the AI SEO agent follows from start to finish, such as a technical SEO audit, run from Ask on the whole site, a segment or URLs.
Snapshot table	An Ask table that holds figures no query over your data can express, such as keyword research results, with up to 50 rows and 8 columns that keep their values.

Edge SEO

Terms for Edge SEO rules, which change the pages bots receive.

Term	Meaning
Blast radius	The number of pages a rule matches, out of all the pages bots requested through EdgeComet in the last 30 days.
Capture Pattern	A regular expression in RE2 syntax, run on the requested URL path, whose numbered groups, $1, $2 and up, fill in a redirect's Destination URL.
Edge SEO	Making SEO changes in the layer between your server and the crawler instead of in your CMS or code, which EdgeComet does by rewriting each page it serves to bots.
Edge SEO rule	A rule that names the pages it applies to and what to change on them, applied when EdgeComet renders or fetches a matching page and before it caches the result.
Edge SEO snippet	A script tag in your page head that applies your live Edge SEO rules in visitors' browsers, since visitors open your server directly and do not get the changes bots get.
Page edit	One action on selected page elements in Edit on the page, such as Rewrite text, Alt text, Move up, Change the tag or Remove.
Response rule	An Edge SEO rule that answers bots with a 301 or 302 redirect, a 404 or a 410 instead of the page.
Rule cascade	The order in which live Edge SEO rules run on a page, the order they went live, where each rule sees earlier edits until a redirect, 404 or 410 rule ends it.
Rule filter	A row of a multiple-pages Edge SEO rule made of a field, an operator and a value, which decides the pages the rule matches.
Rule scope	The choice of which pages an Edge SEO rule changes: One page, which matches one exact address, or Multiple pages, which matches every page its filters match.
Rule status	The state of an Edge SEO rule: Draft (saved, not live), Live (runs on every matching page) or Rolled back (out of production, kept with its history).
Rule variable	A placeholder in braces, such as {h1} or {url_path}, that takes the value of the page an Edge SEO rule is changing.
Structured data at the edge	Edge SEO adding, fixing and removing JSON-LD structured data on the pages search engines and AI bots receive, built from what each page shows.
Value step	One operation in the ordered list a text edit of a multiple-pages rule runs, such as Set, Prepend, Find & Replace or Truncate, each working on the previous step's result.

Configuration

Terms for segments, custom extractions, link regions and the setup EdgeComet runs on its own.

Term	Meaning
Annotation	A dated note on EdgeComet report charts, such as a release, a migration or a Google update, shown as a dot under the date axis for everyone who can open the website.
Archetype	The page type a segment stands for, such as Product pages or Blog articles, picked from a grouped list that starts at Unclassified.
Content link	A link that sits in the Main content or Catalog listing region of a page.
Custom extraction	A named field, such as price, stock or author, that EdgeComet reads from the pages it processes for bots, each one coming from an extraction rule.
Extraction rule	A rule that reads one field from each rendered page with a CSS selector, a regular expression or both, plus an action and a data type.
Extraction rule health	EdgeComet's daily check of each field rule for matching, values and value quality, shown as health labels and followed by automatic repairs of broken rules.
Guidance note	A plain-language instruction that steers every later analysis behind your site's generated crawl budget rules, kept until you delete it.
Link region	A named part of the page where links sit, such as Navigation or Main content, which lets EdgeComet tell content links from template links.
Page chrome	The repeated parts of a page that are not its content, such as a cookie consent bar or a mobile menu, which a page chrome rule removes from the page's main text.
Repair changelog	The log of automatic changes, repairs and improvements to extraction rules, newest first, where Revert puts an earlier version of a rule back live.
Saved segment	A segment from the Segments page as listed under Saved segments in a report Filter, unlike unnamed rules under Filter manually that apply only to the current view.
Segment	A named group of URLs, such as product pages or blog articles, defined by filter rules that must all match, which reports, Data Explorer, alerts and Ask can narrow to.
Segment run	One pass of the AI agent over your site, started by EdgeComet or by you, that proposes segments on the Generating segments run page.
Setup step	One of the six setup steps EdgeComet runs for every website in a fixed order, each starting once the site has the data and the earlier steps it needs.
Template link	A link that sits in any link region other than Main content or Catalog listing, such as Navigation, Footer or Interlinking.
Track changes	A switch on each extraction rule, on by default, that puts the rule's value changes in the Changes report and offers the rule to change alerts.
Zero configuration	EdgeComet configuring itself from your site's own data over its first two weeks: branded keywords, segments, custom extractions, crawl budget rules, link regions and the first Action Board.

Shelf Edge (E-commerce)

Terms for Shelf Edge, the e-commerce reports, labeled E-commerce in the dashboard.

Term	Meaning
Availability tier	One of seven stock states, such as In stock, Out of stock or Withdrawn, that EdgeComet's AI assigns to a product from the availability field on its page.
Catalog fields	Custom extraction rules with fixed names, such as product_name, price and listing_title, each reading one value from the rendered page, which Shelf Edge uses to recognize product and listing pages.
Facet layer	The set of listing URLs that carry a query string, grouped by parameter name such as ?price= or ?p=, on the Shelf Edge Listings report.
Feed check severity	The P0 to P3 level of a Merchant Center feed check, from P0, Google rejects the item or a feed did not import, to P3, hygiene.
Known product URL	A product URL EdgeComet knows of, counted once: a product page in the crawl or a product link in your current Merchant Center feed.
Listing size	A listing page's result_count value, or the number of products shown when it has none, banded from 0-2 products (near-empty) to 50+ products.
Page field change	A change to a feed item field that shows on the product page, such as the title, price or availability, which makes EdgeComet re-cache the page.
Page type (Shelf Edge)	The Archetype set on a segment, such as Product pages or Blog articles, which the Shelf Edge Overview uses as one row of crawl allocation against demand.
Product pages against the feed	Catalog Health crawl checks that compare the price and stock on the product page bots received with your Merchant Center feed, in both directions.
Sellable tier	The availability tiers In stock, Low stock, Made to order and Lead time taken together, as the Changed since last crawl card groups stock moves.
Shelf Edge	The e-commerce module of EdgeComet, labeled E-commerce in the dashboard, that shows how search engines and AI assistants treat an online store's product pages, category pages and Merchant Center feed.
Under-crawled page type	A page type that gets under 1% of Googlebot fetches in the crawl and earns Google impressions.

Data Explorer and Page Analysis

Terms for the row-level tables and the one-URL view.

Term	Meaning
Bot log	The Data Explorer table that lists every request a search engine or AI bot made to your site through EdgeComet, one row per request, with the page the bot received.
Compare window	The second period after vs in the Period row of the Search Console tables and New and lost URLs, by default the same length ending where the main period starts.
Data Explorer	The row-level view of your site's data, such as bot requests, Search Console pages and keywords, crawled pages, sitemap URLs and feed items, as tables you filter, sort and export.
Lost URL	A URL bots requested in the compare window and not in the period, shown with its numbers from the compare window.
New URL	A URL bots requested in the period and not in the compare window, the default set of Bot log > New and lost URLs.
Page Analysis	The EdgeComet view of one URL on one screen: its status and SEO tags, the bots that requested it, its changes, its Google search performance and the copy bots receive.
SEO change marker	A dashed vertical line on a Page Analysis search chart that marks a date the page's SEO tags or status code changed, labeled with the field or the number of changes.

Account and MCP

Terms for accounts, roles and the MCP server.

Term	Meaning
MCP server	The EdgeComet server that connects AI agents such as Claude, ChatGPT, Codex, GitHub Copilot and Cursor to your site's bot requests, Search Console data, crawl, sitemaps and page change log.
Sub account	A normal EdgeComet account that a client owns, created from an agency account, with the agency as Admin on every website in it, including websites added later.
Website role	The access a person has on one website, Read-only, Manager, Admin or Owner, where each role can do everything the role before it can, plus more.

Changes to this list

• 2026-10-26: Rebuilt from every docs page's terms, EdgeComet-specific terms only, grouped by product area.