Answer: For each pull request review, one model turn per PR head grades the diff against the base rubric and matched rule prompts, then calls `record_review` with a risk level; the model never writes to GitHub. The host then recomputes the policy from trusted state and approves or holds accordingly. Status: Confirmed Entry: tool/cursor/agents-and-automation/what-one-review-does Revision: 1 Digest: sha256:a1f57914a470e8e1 (changes whenever the content changes) Timeline: Created: 2026-11-13T03:09:06Z Last changed: 2026-11-13T03:09:06Z Last verified: 2026-11-20T12:34:21Z Next check: 2026-11-21T12:34:21Z Check interval: Daily Expected change: No fixed schedule Data: Field | Value | Source | As of GitHub write restriction | The model never writes to GitHub. | [1] | 2026-11-20 host recompute | The host recomputes the policy from trusted state. | [1] | 2026-11-20 owners requested on hold | up to two owners, never the author | [1] | 2026-11-20 status comment | updates one status comment in place | [1] | 2026-11-20 Sources: [1] Anysphere. "cursor.com/docs/agent-sdk/templates/agentic-owners.md". https://cursor.com/docs/agent-sdk/templates/agentic-owners.md Role: primary. Accessed: 2026-11-20T12:34:21Z Quote: "The model never writes to GitHub." (Supports: GitHub write restriction) Quote: "The host recomputes the policy from trusted state." (Supports: host recompute) Quote: "it requests up to two owners, never the author," (Supports: owners requested on hold) Quote: "and updates one status comment in place." (Supports: status comment) Snapshot SHA-256: 585668d565694eacc9bf097fe85122e3cce0a78b5c2f10dc1ab5d8fc91cf34ec (captured 2026-11-15T12:35:49Z) Verification: Method: Automated extraction from [1] with deepseek-flash (JSON output, thinking disabled). Every value was checked by code to appear verbatim in a quoted passage of [1]. On this check the text of [1] was unchanged since 2026-11-13T03:09:06Z, so the values verified then were kept. Cross-check: not-available at 2026-11-20T12:34:21Z. No independent second source is configured for this entry Last successful check: 2026-11-20T12:34:21Z Last attempt: 2026-11-20T12:34:21Z Notes and limitations: Extracted automatically from a single official source (section "What one review does"). Related: https://latestn.com/tool/cursor/agents-and-automation/when-builds-occur When do Cloud Agent Builds occur in Cursor, including recurring and skipped builds? https://latestn.com/tool/cursor/agents-and-automation/when-to-use-agent-sdk When should I use Cursor Agent SDK instead of other approaches? https://latestn.com/tool/cursor/commands-and-shortcuts/change-pr-parent-stack How do I change or remove a pull request's parent stack in Cursor Origin? https://latestn.com/tool/cursor/agents-and-automation/triage-tracker-template How do I triage a tracker ticket with a Cursor Agent SDK template? https://latestn.com/tool/cursor/agents-and-automation/trigger-manual-review How do I trigger a Bugbot review manually on a pull request? https://latestn.com/tool/cursor/agents-and-automation/webhook-events What events does Cursor send to a cloud agent webhook URL? https://latestn.com/tool/ All entries in AI tools Change log: Time | Event | Change | Source 2026-11-13T03:09:06Z | added | Entry created | — Changes for this entry as a feed: https://latestn.com/tool/cursor/agents-and-automation/what-one-review-does/latest Challenge this entry: Open challenges: None How: If a value here is wrong or out of date, send what is wrong and a link to evidence. Challenges are queued and checked against the evidence; they do not change the entry by themselves. If the entry changes as a result, the change log says so. Form: https://latestn.com/dispute/tool/cursor/agents-and-automation/what-one-review-does POST: https://latestn.com/dispute/tool/cursor/agents-and-automation/what-one-review-does with claim, evidence_url, and optionally field and contact (form-encoded or JSON) GET: https://latestn.com/dispute/tool/cursor/agents-and-automation/what-one-review-does?claim={what is wrong}&evidence_url={link} How to cite: LatestN. (2026). What does the Cursor owners agent do for each pull request review? Revision 1, last verified 2026-11-20T12:34:21Z. https://latestn.com/tool/cursor/agents-and-automation/what-one-review-does (accessed 2026-11-21T10:07:29Z). @misc{LatestN_tool_cursor_agents_and_automation_what_one_review_does_r1, title = {What does the Cursor owners agent do for each pull request review?}, author = {{LatestN}}, year = {2026}, howpublished = {\url{https://latestn.com/tool/cursor/agents-and-automation/what-one-review-does}}, note = {Revision 1; digest sha256:a1f57914a470e8e1; last verified 2026-11-20T12:34:21Z; accessed 2026-11-21T10:07:29Z} } Reuse: License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) Attribution: Copying and adapting are allowed, including commercially, if you credit LatestN and link to the entry you used. Each entry ends with a ready-made citation. URL: https://latestn.com/tool/cursor/agents-and-automation/what-one-review-does