# Agent constantly overshoots simple requests and acts without guardrails

**URL:** <https://forum.cursor.com/t/agent-constantly-overshoots-simple-requests-and-acts-without-guardrails/170002>\
**Category:** Feedback\
**Tags:** auto-run, skills, chat\
**Created:** [August 30, 2026, 6:38pm UTC](https://forum.cursor.com/t/agent-constantly-overshoots-simple-requests-and-acts-without-guardrails/170002 "2026-08-30T18:38:03Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![Jakub\_Pelak](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/jakub_pelak/32/119253_2.png) [@Jakub\_Pelak](https://forum.cursor.com/u/Jakub_Pelak)\
**Post date:** [August 30, 2026, 6:38pm UTC](https://forum.cursor.com/t/agent-constantly-overshoots-simple-requests-and-acts-without-guardrails/170002/1 "2026-08-30T18:38:04Z")

</div>

### Where does the bug appear (feature/product)?

Cursor IDE

### Describe the Bug

The agent routinely does far more than it is asked. Scope should match the request. If I say “build me this app,” I expect it to start building. If I say “change a few files,” I expect only those files to change, then stop. Instead, a small request often turns into a coding spree: extra edits, extra files, staging, even commits I never mentioned.

I should not have to say “do only this, step by step” on ordinary tasks. I often do, because otherwise it runs like a dog that has slipped its leash. That reminder should not be required. The request itself is the scope.

I do want the agent to notice holes in my thinking and correct me. Say so, then wait. Do not “help” by charging ahead into work I did not ask for.

### Steps to Reproduce

One concrete case: I asked to drop a local docs folder and point AGENTS.md at vendor library docs. I did not ask to stage or commit. The agent committed anyway. I did not invoke a commit workflow. A skill file existing on disk is not a request.

### Expected Behavior

Expected: the size of the work matches the size of the ask.  
Actual: one narrow request, then an unsupervised rampage, then I spend the next turn cleaning up.

### Operating System

MacOS

### Version Information

Version: 3.18.9 (Universal)  
VS Code Extension API: 1.128.0  
Commit: 2ba48ff3f7514cc4643c52ca9f7b3173d9b66130  
Date: 2026-08-27T01:42:22.092Z  
Layout: IDE  
Build Type: Stable  
Release Track: Default  
Electron: 40.10.3  
Chromium: 144.0.7559.236  
Node.js: 24.15.0  
V8: 14.4.258.32-electron.0  
xterm.js: 6.1.0-beta.291  
OS: Darwin arm64 25.6.0

### For AI issues: add Request ID with privacy disabled

Request ID: 38cef70d-8c7a-4226-bb77-df05daf4f09b

### Does this stop you from using Cursor

Sometimes - I can sometimes use Cursor

---

<div class="post-metadata">

**Author:** ![izgorodin](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/izgorodin/32/117016_2.png) [@izgorodin](https://forum.cursor.com/u/izgorodin)\
**Post date:** [August 30, 2026, 10:08pm UTC](https://forum.cursor.com/t/agent-constantly-overshoots-simple-requests-and-acts-without-guardrails/170002/5 "2026-08-30T22:08:13Z")

</div>

I treat this as an authorization problem, not a prompting problem: derive a mutation envelope from the request—allowed paths, allowed actions, and whether commits are permitted—and fail closed outside it. A skill may suggest a wider workflow, but widening the envelope should require a new explicit confirmation; its presence on disk must never grant authority.

---

<div class="post-metadata">

**Author:** ![mohitjain](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/mohitjain/32/100229_2.png) [@mohitjain](https://forum.cursor.com/u/mohitjain)\
**Post date:** [August 31, 2026, 5:36am UTC](https://forum.cursor.com/t/agent-constantly-overshoots-simple-requests-and-acts-without-guardrails/170002/7 "2026-08-31T05:36:00Z")

</div>

Hey @Jakub_Pelak and @izgorodin,  
A few guardrails that fix this today:

- Approvals: Settings \> Agents \> Approvals & Execution - use Auto-review or Allowlist (not Run Everything). There, `git commit` needs your approval by default. On Auto-review you can also add “every git commit or stage should require approval.” [Run Modes](https://cursor.com/docs/agent/run-modes)
- That commit skill: add `disable-model-invocation: true` to its frontmatter (or remove it) so it only runs when you type `/skill-name`, never on its own. [Agent Skills](https://cursor.com/docs/context/skills)
- Persistent scope: an Always-Apply rule or AGENTS.md line: “Only change the files I ask about. Never stage or commit unless I explicitly ask.” [Rules](https://cursor.com/docs/context/rules)
- Proposal-first: Ask mode is read-only and Plan stays read-only until you approve the plan (Shift+Tab to switch). [Modes](https://cursor.com/docs/agent/modes)

The approval setting plus disabling that skill should stop the unrequested commits. Passing the broader scope feedback along too - if it still overshoots with those in place, send the request ID and I’ll dig in.

---

<div class="post-metadata">

**Author:** ![system](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/system/32/99069_2.png) [@system](https://forum.cursor.com/u/system)\
**Post date:** [September 30, 2026, 12:04pm UTC](https://forum.cursor.com/t/agent-constantly-overshoots-simple-requests-and-acts-without-guardrails/170002/9 "2026-09-30T12:04:21Z")

</div>

Closed on September 30, 2026, 12:04pm UTC

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.
