> ## Documentation Index > Fetch the complete documentation index at: https://docs.attensira.com/llms.txt > Use this file to discover all available pages before exploring further. # How we read communities > Which community threads Attensira reads, how it reads each site (Reddit only through its official Data API, the others only where robots.txt allows), what it keeps, how fast deleted Reddit content is deleted here, and what it never does. Discussion threads on Reddit, Hacker News, Stack Overflow, the Stack Exchange sites and Quora are among the pages AI answers cite most. Attensira reads them to learn why an answer says what it says, to draft a reply you post yourself, and to turn what buyers ask into changes on your own site. This page is the rules for the reading. What happens with what is read is on [Community answers](https://docs.attensira.com/agent/community-answers.md). ## Read, never written Attensira reads communities. It never posts, comments, votes or sends a message on one, and never creates an account there. No tool, route, background job or setting can, in any [approval mode](https://docs.attensira.com/agent/approval.md) or autonomy level. A draft reply is yours to read, edit and post from your own account. ## Which threads are read Only threads that matter to your answers. The thread list is built from the [deciding sources](https://docs.attensira.com/measure/deciding-sources.md): the pages AI answers cite for the questions you track. Building it reads nothing from any community; it uses the answers Attensira already holds. A thread is read when the agent drafts a reply for it or records what it says, and only if it is a thread on one of these sites: * **Reddit** * **Hacker News** * **Stack Overflow and the Stack Exchange sites** * **Quora** A subreddit front page, a user profile or a search page is not a thread and is not read. ## How each site is read | Site | How | Never | | - | - | - | | Reddit | Only through Reddit's official Data API, as an application registered under Reddit's Data API terms. Every request is authenticated as that application and kept under Reddit's published rate limit. | Logged in as you or anyone else. A request to reddit.com outside the API: the page reader refuses every reddit.com address. | | Hacker News, Stack Overflow, Stack Exchange, Quora | Through Attensira's page reader, after reading the site's robots.txt. A path robots.txt does not allow is not read, and is not handed to another reader to get around it. At most one read every two seconds per site from each Attensira server. The robots.txt check never follows a redirect to another site; the page read itself follows redirects as a browser does. | Logged in, with a cookie or a password. | When a site is busy or a limit is reached, the agent is told to try again shortly or move on. A read is never retried in a loop. ### Reddit today Reddit's Data API is not connected to Attensira today, so **no Reddit thread is read**. A Reddit thread still appears in your thread list, with its link and the answers that cite it, because the list comes from those answers. A draft for it can still be written, without reading it, and it says so: **written without reading the thread**. Read the thread and the subreddit's rules before you post. No insight is recorded from a Reddit thread. Reddit is never read some other way instead: not through a scraper, a search engine's copy, or a browser. ## What is kept The agent reads a thread in full while it works, for that run only. The run's record keeps the thread's link, not its text. From a Reddit thread, Attensira keeps: * the thread's id and link; * what it worked out about the thread: its score and number of comments, whether it still takes replies, and whether the subreddit has a self-promotion rule; * short quotes: the title and up to three top comments, each cut to 200 characters, each tied to the post or comment it came from, with no username. Never the post's full text, and never a comment that was already deleted or removed. ### Deleted on Reddit, deleted here within 48 hours Once a day, Attensira asks Reddit about every thread and comment it keeps anything of. * A thread Reddit reports deleted or removed is deleted from every workspace that read it, with the insights recorded from it. * A comment Reddit reports deleted or removed loses its quote. * A quote Reddit has not confirmed for 20 hours, because Reddit could not be asked, is deleted at the next daily check anyway. A deletion on Reddit is honoured within 48 hours either way. A thread's id, link and the facts worked out about it are not Reddit content and stay. Drafts in [Work](https://docs.attensira.com/agent/inbox.md) are yours: the daily check does not edit them. ## What reading is used for * **Drafts.** A reply you post yourself, held to the [rules every draft follows](https://docs.attensira.com/agent/community-answers.md#the-rules-every-draft-follows): disclosure when it names your product, one link at most, the subreddit's rules first. * **Insights.** What buyers ask, what puts them off, and what they believe about you, each filed as a change to your own site or a [Brand Book](https://docs.attensira.com/brand-book/entity-sync.md) proposal for you to approve. See [What the threads tell us](https://docs.attensira.com/agent/community-answers.md#what-the-threads-tell-us). ## Over MCP and the API Reading is the agent's work, so there is no verb or route that reads a community for you. What it found is readable on every door: the thread list with `analyze(view=community_threads)` over MCP or `GET /v1/views/community_threads` over REST, and the insights as listed under [Over MCP and the API](https://docs.attensira.com/agent/community-answers.md#over-mcp-and-the-api). There is no verb that posts, and none will be added. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.