Announced August 13, 2026, X expanded the open-source code behind its For You timeline on GitHub and began testing a tool that lets eligible users see — and download — the labels applied to their account and posts over the prior month. It is the closest X has come to letting creators check for themselves whether their reach is being limited.
2026-08-15 · by Moe Ameen
On August 13, 2026, X expanded the open-source code behind its For You timeline on GitHub and announced a new in-app tool aimed at the platform's most persistent complaint: that reach is throttled quietly, with no way to confirm it. The company framed the two pieces as fitting together — the published code shows how the For You feed is generated, including which content labels can reduce a post's visibility and by how much, while a new "Under the Hood" page shows a user whether those labels are actually applied to their own account or posts. The stated goal, in X's words, is to let "people be able to answer for themselves whether a platform is limiting their reach."
X has long avoided the word "shadowban." Its enforcement philosophy, published in 2023 as "Freedom of Speech, Not Reach," describes reducing a post's distribution — excluding it from search, trends, recommendations, and the For You and Following timelines — as a middle path between leaving content fully up and removing it. What was missing was a reliable way for a creator to see when a reach-limiting label had been applied. The Under the Hood tool is X's answer: eligible users can view the labels on their account and posts from the prior month and download the underlying data.
The rollout is a limited test, not a universal switch. Reporting on the launch indicates the tool is going first to a randomized group of accounts that are at least one year old and have posted enough in the prior month (around ten or more posts) to have data worth showing. X also noted that the code powering some safety decisions — for example, systems that use Grok to predict rule violations — was deliberately excluded from the open-source release so bad actors cannot reverse-engineer and game it. Treat the eligibility thresholds and exact feature naming as a launch-day snapshot and confirm against X's own Help Center and the in-app page, which are the authoritative sources as the test widens.
The honest lesson from X publishing its ranking code is not "learn the labels and dodge them" — X kept the safety-critical logic closed precisely so it can't be gamed. The lesson is that any single feed can quietly cap your reach on rules you'll never fully see, so the only reliable hedge is to not let one algorithm be the gate. That is the exact problem [Kompozy](/) is built around. You create one idea and Kompozy fans it across the eight social platforms plus blog and email, so an X visibility label throttles one channel while the same post keeps compounding on Instagram, TikTok, YouTube, LinkedIn, and the rest — [multi-platform distribution](/glossary/content-repurposing) as an insurance policy against any one platform's opaque ranking.
There is a second, concrete edge. Independent analyses of X's ranking have found that posts carrying links off-platform tend to get demoted, so native content wins the feed. Kompozy is a full generation-and-publishing engine, not a link-in-bio relay — it produces platform-native assets ([Persona Shorts](/glossary/persona-shorts) talking-head video, brand-exact carousels, [Persona Frames](/glossary/persona-frames) composites, quote cards, and text posts across [18 formats](/glossary/output-buckets)) and publishes them straight into each feed on [Autopilot](/glossary/autopilot), no external URL required to earn reach. When you can't see every rule that governs your visibility, the winning move is to ship consistent, original, native content everywhere at once — and that is the workflow Kompozy runs from a single review gate.
Sort of. On August 13, 2026, X began testing an "Under the Hood" page that shows eligible users the visibility labels applied to their account and posts over the prior month, with an option to download the data. It is the first first-party readout of reach-limiting labels, but it is a randomized test — gated to older, active accounts — not a universal feature yet, and X still avoids the term "shadowban."
At launch it rolled out to a randomized test group. Reporting on the release indicates eligibility is limited to accounts at least one year old that have posted actively — around ten or more times — in the prior month, so there is label data worth displaying. Thresholds and naming may change as the test widens; check X's Help Center and the in-app page for the current rules.
It is the enforcement philosophy X published in 2023: instead of removing a rule-bending post, X may reduce its distribution — excluding it from search, trends, recommendations, and the For You and Following timelines while leaving it visible on the author's profile. The 2026 open-source code and Under the Hood tool make which posts carry those reach-limiting labels more visible to the author.
Two moves. First, publish native content — X consistently demotes posts that link off-platform, so keep the value in the post itself. Second, do not depend on one feed: publish the same idea across multiple platforms so a single X label caps one channel, not your whole distribution. A content engine like Kompozy generates platform-native posts and publishes them across the eight social platforms plus blog and email from one workflow.