Every tool your AI gets

Start here. One call shows where everything stands: the plan, every server (online or not, root or limited access, safety mode, paused, its sites), the open problems and warnings AskYourStack's own monitoring found (disk, memory, failed services, SSL certificates, a site not...

List the servers on this account: online or not, safety mode, paused or not.

Step-by-step guide for a common job, written for you to follow. ); seo-dashboard (A site's search performance over the last 90 days as one visual page set in the Inter typeface. )

Remember something for future chats about a server (or the whole account): what is special about it, how it is set up, what must not be touched, a decision the user made.

Delete a saved note that is wrong or no longer true (its id is in overview).

Write down finished work, when a task that changed something is done: two or three plain lines (up to 800 characters) a non-technical owner understands. What you did, what changed, what is left for them.

Distro, kernel, CPU, memory, disk, package manager, running services, listening ports and any Magento or WordPress installs found. Call this first on a server.

Run a shell command as root (bash -c). No terminal, no stdin. Returns exit code, stdout and stderr (long output keeps the first 16 KB and last 48 KB). Destructive commands need the user's approval.

Start a long command (installs, composer, compiles, imports, backups) in the background as root, in its own systemd unit, so it keeps running after this call returns and through agent updates.

Status (running, finished, ended), exit code and output of a background job. Without offset you get the last 48 KB; pass next_offset from the previous answer to get only what is new.

Stop a running background job and everything it started.

The 20 most recent background jobs on the server with their status.

Health of the server now: disk and inode use per filesystem, free memory, swap, load per CPU, failed services, SSL certificates and days left, pending security updates, whether a reboot is needed, backups (how old the newest is, the rhythm of the last runs, backup programs that...

Read a log by name for a time window, with repeated messages counted once: how many different errors there are, how often each happened, when first and last, a sample of each (with the top of its stack trace) and the latest lines.

Read a file (up to 1 MB). Binary files come back base64.

Create or replace a file. Keeps owner and mode of an existing file and saves its previous version under /var/lib/askyourstack/backups. Parent folders are created.

Save folders (and optionally MySQL/MariaDB databases, dumped with mysqldump) on the server so they can be restored with rollback. Refuses when it would use more than half the free disk.

Snapshots on the server, newest first, with what each holds.

Write a snapshot's files back (all of it, or only some paths). Saves the current state first, so a rollback can itself be undone. Files created after the snapshot are left alone; database dumps are not loaded (use run).

When a job has several risky steps (deleting folders, dropping tables, changing SSH or closing access in the firewall, a reboot), list the exact calls you mean to make and ask for the user's approval once, instead of stopping at each one.

Wait for the user to decide, so they do not have to come back to the chat and tell you. Right after you give the user an approval link (approval_id), a plan link (plan_id), the link to open changes on a server (server) or a setup link while nothing is connected yet (setup)...

The sites on the server: Magento 2, OpenMage (Magento 1), WordPress and WooCommerce, PrestaShop, Shopware 6, Drupal, Joomla, OpenCart, Laravel and Ghost, wherever the hosting panel keeps them (found from the web server's own configuration too) and inside Docker containers when...

Run one SQL statement on a site's database, connecting with the site's own credentials (you never need the password).

Run bin/magento as the site's owner in its root, like args ["cache:flush"] or ["indexer:reindex", "catalogsearch_fulltext"].

Run wp-cli as the site's owner with --path set, like args ["plugin", "list"]. Listing and checksum commands run straight away; deleting plugins or users, db reset/import, search-replace without --dry-run and eval need approval.

php for Joomla, ghost (ghost-cli) for Ghost (list_sites shows each site's console; Magento 2 has the magento tool and WordPress the wp tool). Pass the command and its options as args, like ["cache:clear"] or ["plugin:list", "--json"].

Quick malware check of a site (most thorough for Magento and WordPress; for the other kinds list_sites shows it checks the files, upload folders, cron and processes, not the database): backdoor patterns in PHP, PHP in upload folders, disguised and hidden PHP, recently changed...

List the SEO titles, meta descriptions or image alt texts of a WordPress site (Yoast SEO, Rank Math, SEOPress or The SEO Framework) or a Magento 2 shop (products, categories, CMS pages and product images at the default store scope), with their current values.

Set SEO titles, meta descriptions or image alt texts on a WordPress site or a Magento 2 shop, up to 30 fields a call, by the ids seo_fields_list gave.

Put back the values a seo_fields_apply batch replaced (kept 90 days). batch is the id from seo_fields_apply or seo_fields_history.

The batches of SEO title, description and alt text changes made on a site in the last 90 days, with what each changed and whether it was undone. Changes nothing.

Find the paths of a site on the user's server that answer 404: from the crawler hits in its access logs (the last hours, default 168) and, when Search Console is connected, pages that had clicks in the last 90 days and are gone now.

The custom redirects a site holds: Magento's url_rewrite table, or WordPress with the Redirection or Rank Math plugin. Changes nothing.

Write redirects (from a path, to a path or full address, code 301 or 302) into the site's redirect store: Magento url_rewrite, or the WordPress Redirection or Rank Math plugin.

Remove the redirects one redirects_apply batch wrote. batch is the id from redirects_apply or redirects_history.

The batches of redirects written on a site in the last 90 days, each with its redirects and whether it was undone. Changes nothing.

gz file on its server, with the site's own credentials (they never reach the chat). Runs in the background: returns job_id and the file path; follow it with job_output.

webp, with the encoder the server has (cwebp, avifenc, ImageMagick or vips; the answer names it, or says what to install).

). Each with its size and age. Use it to find what db_restore can load back.

gz dump into a site's database, with the site's own credentials: the undo for a bad change, or a restore from a backup. gz file on the server (a backup the user keeps). ), so the restore itself can be undone with the same tool.

Copy a folder from one of the user's servers to another, directly between them, for migrations (rsync over SSH with a temporary key that only allows reading that folder; nothing goes through AskYourStack).

Close the copy links between servers (removes the temporary keys). Call it when a migration is done. Without from/to, closes every open link.

The Google Search Console properties the user picked for AskYourStack, with the access level, and which Google account is connected. Call this first for SEO work that uses Search Console.

Clicks, impressions, CTR and position against the period before, with the biggest winning and losing queries and pages

What to work on first, ranked by clicks at stake: striking-distance queries, page-one pages with a low CTR, decaying pages and competing pages

Performance by query, page, country, device or date, with filters and two-period comparisons

Google's own view of up to 20 URLs: indexed or not and why, last crawl, the canonical Google picked

Submitted sitemaps with errors, warnings and when Google last read them

Queries where two or more of your pages split the impressions and compete with each other

Google's core, spam and helpful content updates, to line up with traffic changes

The Google Merchant Center accounts the user's connected Google account can see (sub-accounts of a multi-client account included), with their ids. Call this first for Shopping and free listing questions.

Why products are not showing in Google Shopping ads and free listings: per destination and country the number of products active, pending, disapproved and expiring, the account's own issues (suspensions, missing settings, website claims), and the product issues grouped by kind...

Products in Merchant Center with their issues: offer id, title, link, and each issue with severity, the attribute concerned, where it applies and Google's explanation.

The Google Analytics 4 properties the user picked for AskYourStack, each with a health check: time zone, currency, data retention, key events (conversions) set up, and whether data is still coming in (sessions yesterday and in the last 7 days).

Sessions, users, engagement, conversions and revenue against the period before, by channel, with organic search's share

Landing pages with engagement, conversions and revenue, for organic search or any channel

Visits from ChatGPT, Perplexity, Gemini, Claude, Copilot and other AI assistants, and the pages they land on

Any GA4 report: pick dimensions, metrics, filters and a comparison period

Who is on the site right now, to check a launch, a campaign or that tracking works

Tells Bing, Yandex and the other IndexNow search engines which pages are new, changed or gone, so they crawl them now

The sites in the user's Bing Webmaster Tools account and which of them the AI may read (picked on the SEO page).

Bing's clicks and impressions by day, the last 28 days against the 28 before, top queries and pages

What Bingbot crawled and indexed, its errors, and the crawl issues per address

Bing's own search counts for a keyword, its monthly history and related keywords, free

Submit pages or a sitemap to Bing, with the quota left

The sitemaps Bing knows for a site, with status, when Bing last read them, and how many addresses they hold. To submit one, use bing_submit with sitemap.

Inbound links Bing knows for your pages, with anchor text (Search Console has no links API)

Google's monthly search volume, 12-month history, CPC and competition for up to 200 keywords, in any country and language

Keyword suggestions around a seed, with volume, keyword difficulty and search intent

The live Google results for a keyword, desktop or mobile, with your site's positions and the page features (AI overview, snippets, people also ask)

What any domain ranks for, with position, volume, estimated visits and page, and the domains competing with it

What Googlebot, Bingbot, AI crawlers and SEO bots really fetched, from your access logs: errors, crawl waste, fake Googlebots

Structured data checked against schema.org and Google's rich-result requirements

Real-user LCP, INP and CLS from the Chrome UX Report, 6 months of history, and a Lighthouse test with the causes

Status, redirect chains, titles, descriptions, canonicals, hreflang, headings, robots and issues for up to 10 URLs

Every link on a page with anchor text and rel, and broken or redirecting internal links

A page's main content as clean text, to review, compare with a competitor or rewrite

robots.txt tested the way Google reads it, for Googlebot, Bingbot and AI crawlers

An llms.txt for your site, the file AI assistants read: name, summary and key pages from your sitemap with their own titles and descriptions

Sitemaps audited: limits, lastmod, and a sample of URLs that should answer 200 and be their own canonical

A crawl of your whole site from your own server (up to 5,000 pages, no CDN in the way): broken links, redirect chains, repeated titles, thin and duplicate pages, sitemap problems, read by issue with site_audit_report

Read a site audit that site_audit made. Without more arguments: the summary of the latest run (pages by status and by clicks from the start page, every issue with its count, why it matters and three examples, and what changed since the run before).

Daily Search Console, Analytics and Core Web Vitals figures kept for as long as you like, and alerts when clicks fall, a top page leaves the index or a vital turns poor

Writes the disavow file for Google from your Ahrefs, Semrush, SE Ranking or Bing link data, merged with the one Google already has, ready to upload

Sites the user connected without a server: a WordPress site through its REST API (any host, no SSH) or a WhizzyCommerce shop through the shop's own connector.

List content of a connected site, newest change first. WordPress: posts, pages, media, categories, tags and the site's own content types (products on a WooCommerce shop, for example).

Read one item of a connected site in full, as its editor would see it: title, content (the stored source), status, address and what else it has.

Create or change content on a connected site. Without id it creates (as a draft unless status says otherwise); with id it changes only the fields you pass.

Add an image to a connected site's media library (WordPress) or files (Shopify) from a public address (jpeg, png, gif, webp or avif, up to 15 MB), with its alt text and title.

Move a post, page or other item of a WordPress site to its trash (WordPress keeps it 30 days; content_undo brings it back), or remove a URL redirect of a Shopify store (content_undo makes it again).

Undo a change made through content_save, content_upload or content_trash on a connected site: an edit gets its old values back, a created item goes to the trash, a trashed one comes back.

For a connected WhizzyCommerce shop (see connected_sites): what the shop lets an AI do right now, as its own list of capabilities with their arguments: orders and shipping, products, categories and imports, prices, sales, coupons and gift cards, customers, languages and...

Run one capability of a connected WhizzyCommerce shop, by the name whizzy_tools gave, with its arguments. Reads answer straight away.

The Cloudflare zones (domains) the user connected, with status, plan, name servers and what the AI may do there. Start here before any other cf_ tool. Cloudflare is connected on the Sites & Apps page with an API token; no server is needed.

List the DNS records of a Cloudflare zone, filtered by type, name or search (part of a name or content), each with its id. Changes nothing.

Add a DNS record to a Cloudflare zone: a new subdomain, a mail record or a verification text. proxied defaults to false. It can be removed again with cf_undo.

Change one DNS record of a Cloudflare zone, by record_id from cf_dns_list (or name plus type when they match one record): its content, TTL, proxy, priority or name.

Remove one DNS record of a Cloudflare zone, by record_id from cf_dns_list (or name plus type when they match one record). Records that were there before AskYourStack need the connection's "change anything" mode. cf_undo creates it again.

Clear Cloudflare's cache of a zone: everything, or only the given addresses (up to 30). Use it after deploying changes a visitor still sees old, or a wrong page cached. Nothing to undo; the cache fills again from the origin.

List the custom firewall rules and rate limits of a Cloudflare zone, each with its id, action and expression. Changes nothing.

Add one firewall rule or rate limit to a Cloudflare zone. What it matches comes from ip or ips (addresses or CIDR), asn, country or countries, path (starts with), user_agent (contains), host and not_verified_bot, combined with and; or a raw expression in Cloudflare's rule language.

Remove one custom firewall rule or rate limit from a Cloudflare zone, by rule_id from cf_firewall_list. cf_undo adds it again.

Read or change a Cloudflare zone's settings: security_level (essentially_off, low, medium, high, under_attack) or under_attack true/false as a shortcut, development_mode (skip the cache for 3 hours), browser_check, automatic_https_rewrites; and in "change anything" mode ssl...

Traffic through Cloudflare for a zone (proxied traffic only): requests, cached share, visitors, threats, bandwidth and status codes per day for up to 30 days, top countries, requests per hour for the last 24 hours, and a sample of the firewall events of the last 24 hours (what...

Undo a Cloudflare change made through the cf_dns_, cf_firewall_ or cf_settings tools: a record or rule that was added is removed, a changed one gets its old values, a removed one is made again.