---
id: "CVE-2026-95985"
kind: "vulnerability"
aliases: ["CVE-2026-95985"]
title: "Kiro IDE: file write tool lets crafted repos run commands and poison agent"
severity: "high"
cvss: 8.8
epss: 0.00135
exploited: false
status: "open"
category: "ai-tooling"
products: []
ecosystems: []
tags: ["agentic-ide", "ai-tooling", "aws", "desktop", "developers", "ide", "kiro", "lab-disclosure"]
not_affected_if: ["kiro-1.0.242-or-later", "no-kiro-installed"]
published_at: "2026-09-24T00:00:00+00:00"
first_seen_at: "2026-10-24T20:05:36+00:00"
updated_at: "2026-10-25T12:38:49+00:00"
sources: [{"source": "lab", "url": "https://aws.amazon.com/security/security-bulletins/rss/2026-117-aws/", "first_seen_at": "2026-10-24T20:05:36+00:00"}, {"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-95985", "first_seen_at": "2026-10-24T20:15:27+00:00"}, {"source": "epss", "url": "https://api.first.org/data/v1/epss?cve=CVE-2026-95985", "first_seen_at": "2026-10-25T12:38:49+00:00"}]
---

# Kiro IDE: file write tool lets crafted repos run commands and poison agent

Kiro IDE, an agentic coding editor, has a file write tool that, when the agent runs in a crafted untrusted repository, can modify auto-loaded global configuration paths on the user's machine. An attacker who supplies such a repository can get arbitrary command execution and inject instructions into the agent's context.

_Summary, check and fix written by a model from the sources; labelled as such._

## How to check

- Check the installed Kiro IDE version and compare it against `1.0.242`.
- Review the global Kiro configuration directory `~/.kiro` on macOS and Linux, or `%USERPROFILE%\.kiro` on Windows, for entries the user did not create.

## Fix

- Upgrade Kiro IDE to `1.0.242` or later.
- For anyone who ran the agent in an untrusted workspace on an earlier version, review `~/.kiro` (or `%USERPROFILE%\.kiro`) for unknown entries and remove them.

## Timeline

- 2026-10-24T20:05:36+00:00 lab https://aws.amazon.com/security/security-bulletins/rss/2026-117-aws/ — AWS: CVE-2026-95985 - Kiro IDE Allows Agentic Writes to Global Configurations While Working in Untrusted Workspaces
- 2026-10-24T20:15:27+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-95985
- 2026-10-25T12:38:49+00:00 epss https://api.first.org/data/v1/epss?cve=CVE-2026-95985

## Documents

- [advisory] AWS: CVE-2026-95985 - Kiro IDE Allows Agentic Writes to Global Configurations While Working in Untrusted Workspaces https://aws.amazon.com/security/security-bulletins/rss/2026-117-aws/

## From the sources

CVE-2026-95985 - Kiro IDE Allows Agentic Writes to Global Configurations While Working in Untrusted Workspaces

### CVE-2026-95985 - Kiro IDE Allows Agentic Writes to Global Configurations While Working in Untrusted Workspaces

Bulletin ID: 2026-117-AWS Scope: AWS Content Type: Important (requires attention) Publication Date: 09/24/2026 10:00 AM PDT

Description:

Kiro is an agentic IDE that users install on their desktop. We identified CVE-2026-95985. The file write tool in Kiro IDE before version 1.0.242 might allow remote unauthenticated actors to execute arbitrary commands and to inject crafted instructions into the agent's context. When a user runs the agent in a crafted repository as an untrusted workspace, sending any message can cause agent modifications to auto-loaded global configuration paths.

Impacted versions: < 1.0.242

Resolution:

This issue has been addressed in Kiro IDE version 1.0.242. We recommend upgrading to the latest version. Users who ran the agent in an untrusted workspace on an earlier version should also review the global Kiro configuration directory (~/.kiro on macOS and Linux, %USERPROFILE%\.kiro on Windows) for entries they did not create.

Workarounds:

None.

References:

- CVE-2026-95985

Acknowledgement:

We would like to thank Dinh Ngoc Dung for collaborating on this issue through the coordinated disclosure process.

 Please email aws-security@amazon.com with any security questions or concerns.

---

_Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai_
