X search operator, decoded

from:anthropicai since:2026-06-01what the query does, and what it returns

This is an X advanced-search query, not an error message and not a product name. It asks X for every post by the official @AnthropicAI account dated on or after June 1, 2026. Below: the exact mechanics, a verified list of what actually showed up, and a way to keep the query running without refreshing it by hand.

M
Matthew Diakonov
6 min read

Direct answer · verified June 23, 2026

from:anthropicai since:2026-06-01 returns every post from @AnthropicAI on or after June 1, 2026. You must be signed in to X for it to work. As of June 23, the two posts that dominate that window are Claude Fable 5 going generally available on June 9 and the June 12 statement that a US government export-control directive forced Anthropic to disable Fable 5 and Mythos 5 for all customers (both still offline as of June 22). Around them: enterprise-managed MCP connectors, self-hosted Managed Agent sandboxes, and Claude Platform on AWS.

Run it yourself: @AnthropicAI on X · anthropic.com/news

How the query is built

The string is two operators with an implicit AND between them. X evaluates both and returns the intersection. Most guides stop at "type it in the box," but the parts that trip people up are the date format and the boundary behavior, so here is the precise version.

from:anthropicai# posts authored by @AnthropicAI
since:2026-06-01# on or after June 1 (inclusive)
x.com/search?q=from%3Aanthropicai%20since%3A2026-06-01

Rules that actually matter

  • Dates are YYYY-MM-DD only. 06/01/2026 or 1-6-2026 silently fail.
  • No space after the colon. from: anthropicai with a space is treated as a keyword, not an operator.
  • since: is inclusive (includes June 1); until: is exclusive (until:2026-07-01 stops at June 30).
  • You must be signed in. Logged out, the query bounces to a login wall and returns nothing.
  • Sort by the Latest tab. The default Top tab hides chronological order, which defeats a date-bounded query.

The full operator set lives in X's own advanced-search help, or build the query visually at x.com/search-advanced.

What the query returns right now

If you cannot run it at this moment, here is the substance of the window, verified on June 23 against the account and Anthropic's newsroom. The feed is unusual this month: a flagship launch and an abrupt government-ordered shutdown landed three days apart.

Claude Fable 5 generally available

Jun 9, 2026

The Mythos-class model went GA, positioned above Opus 4.8 with a focus on long-context work across millions of tokens and front-line coding evals. This is the post most people are looking for when they run the query.

anthropic.com/news

US export-control directive, Fable 5 and Mythos 5 disabled

Jun 12, 2026

Three days after the launch, Anthropic said it received a US government export-control directive (delivered 5:21pm ET) ordering it to suspend access to Fable 5 and Mythos 5 for all foreign nationals. To comply, it disabled both models for every customer. Access to all other Claude models was unaffected. As of June 22 both remained offline.

anthropic.com statement

Enterprise-managed MCP connectors (beta)

Jun 2026

Admins on Team and Enterprise can provision MCP connectors once, starting with Okta, so users get zero-touch access on first login, with centralized authorization across Claude chat, Claude Code, and Cowork.

anthropic.com/news

Self-hosted Managed Agent sandboxes (public beta)

Jun 2026

Claude Managed Agents can now run tool execution in an environment you control, your own infrastructure or a managed provider, and connect to your private MCP servers.

anthropic.com/news

Claude Platform on AWS

Jun 2026

The Claude API became reachable through native AWS endpoints with AWS billing and IAM authentication, covering Messages, Files, Batches, Managed Agents, Agent Skills, code execution, and tool use.

anthropic.com/news

Dates and details cross-checked against Anthropic's newsroom and contemporaneous reporting; the export-control shutdown was confirmed on Anthropic's own statement page.

The annoying part: this query goes stale the moment you close the tab

A one-time lookup is easy. Tracking @AnthropicAI over weeks is the actual job, and it is tedious for a specific reason: X gates search behind login. Headless scrapers hit the login wall, so the "read tweets without an account" mirrors that show up in results are usually hours or days behind, and they break whenever X tweaks its markup. The reliable way to read your own logged-in feed is to use a browser that is already signed in as you.

That is the concrete reason a desktop agent can do this where a cloud scraper cannot. Fazm wraps the Claude Code agent loop in a native Mac app, and its browser automation drives your actual Chrome through an extension, not a throwaway headless instance. A line from the app's changelog (version 2.9.71, June 21, 2026) makes the behavior explicit: "Agent now opens authenticated sites (banking, dashboards) in your Chrome instead of the system default browser, so you stay logged in." The same property that lets it reach your bank dashboard lets it reach your signed-in X search. You never paste a cookie or a password into a third party; the session is the one already on your machine.

What the agent does with the query

01 / 05

1. Open your Chrome

The agent attaches to the Chrome you already use, with your X session live. No separate login, no credential handoff.

Reading the query manually vs. in your own session

FeatureHeadless scraper / logged-out mirrorFazm in your Chrome
Gets past X's login wallNo, search needs a signed-in sessionYes, uses your already-signed-in Chrome
Where your credentials goPasted into a third-party tool, or scraped blindNowhere; the session stays on your machine
FreshnessOften hours or days behind, or emptyLive, the same results you would see
Reads the results asBrittle HTML scraping that breaks on redesignsStructured content via accessibility APIs
Survives a restartRe-run from scratch each timePersistent session, the watch resumes

Fazm is a macOS app (14.0+) that runs locally and uses your own Claude account. It is not an X scraping service; it just drives the browser you already use.

Why this particular feed is worth watching if you run Claude Code

For most people, @AnthropicAI is news. If you run Claude Code through a wrapper on your own account, it is closer to a status page. The agent loop in the app is the real Claude Code, so a model going GA or getting pulled is not a headline you read later, it is a change to the tool you are mid-task in. The June 12 directive is the cleanest illustration: a model that was generally available on June 9 was disabled for every customer by June 12.

That is why the wrapper Fazm bundles Codex as a swappable backend per chat and supports custom API endpoints. If one model disappears between sessions, you switch the backend instead of losing the window. Watching the feed and being able to act on it are two halves of the same workflow, which is the whole reason a tracking query like this one is worth automating.

Want an agent watching the feeds you actually care about?

Book 20 minutes to talk through running Claude Code on macOS with a browser agent that reads your own logged-in sessions.

from:anthropicai since:2026-06-01, common questions

What does from:anthropicai since:2026-06-01 actually do?

It is an X (Twitter) advanced-search query. from:anthropicai restricts results to posts authored by the @AnthropicAI account. since:2026-06-01 restricts them to posts created on or after June 1, 2026. Put together, you get every post from Anthropic's official account from June 1 onward, newest first when you sort by Latest. The date must be YYYY-MM-DD and there is no space after either colon.

Is since: inclusive or exclusive?

since: is inclusive, so since:2026-06-01 includes posts made on June 1 itself. Its partner until: is exclusive, so until:2026-07-01 stops at June 30 and does not include July 1. If you want a clean June-only window, use since:2026-06-01 until:2026-07-01.

Why does the search return nothing or send me to a login screen?

X requires you to be signed in to run advanced search. Logged out, the query URL bounces to a login wall and returns no results. The same is true for headless scrapers and most third-party 'view tweets without an account' mirrors, which is why they are often stale or empty. Run it in a browser where you are already signed in.

What were the headline posts from @AnthropicAI since June 1, 2026?

Two dominate the feed: Claude Fable 5 going generally available on June 9, and the June 12 statement that a US government export-control directive forced Anthropic to disable Fable 5 and Mythos 5 for all customers (still offline as of June 22). Around them sit platform posts: enterprise-managed MCP connectors, self-hosted Managed Agent sandboxes, and Claude Platform on AWS.

How do I open the query directly instead of clicking through Advanced Search?

Paste the query into the search box at x.com/search-advanced, or build the URL yourself: x.com/search?q=from%3Aanthropicai%20since%3A2026-06-01 then click the Latest tab. The %3A is the URL-encoded colon and %20 is the space between the two operators.

Can I have an agent watch this query for me instead of refreshing it by hand?

Yes, if the agent drives a browser where you are already signed in. Fazm's agent opens authenticated sites in your own Chrome through its extension rather than a headless instance, so it can load the query, read the Latest results, and summarize new posts without you handing X credentials to a scraper. You set the cadence; it reports what changed.

Why does Anthropic's June feed matter if I use a Claude Code wrapper?

Because the wrapper runs the real Claude Code agent loop against your own Claude account, model and access changes land on you directly. The June 12 shutdown of Fable 5 and Mythos 5 is the clearest example: a model can disappear between sessions. Fazm bundles Codex as a swappable backend and supports custom API endpoints, so a single model being pulled does not strand you.

How did this page land for you?

React to reveal totals

Comments ()

Leave a comment to see what others are saying.

Public and anonymous. No signup.