Skip to main content
Vulnerability Database/CVE-2026-111079

CVE-2026-111079: Goose Information Disclosure Vulnerability

CVE-2026-111079 is an information disclosure flaw in Goose that exposes sensitive terminal data to LLM providers. This post explains its impact, affected versions, and mitigation steps.

Published:

Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

Sources: National Vulnerability Database, CISA KEV, Exploit-DB

CVE-2026-111079 Overview

CVE-2026-111079 is an information disclosure vulnerability in Goose. When the terminal integration feature is enabled, the application captures shell command history and forwards it verbatim to configured Large Language Model (LLM) provider endpoints. The transmitted content is sent without sanitization or redaction. Users who enter sensitive data in the terminal, such as API tokens, passwords, or environment variables containing secrets, risk exposing those values to third-party LLM providers and their logs. The flaw is classified under CWE-497: Exposure of Sensitive System Information to an Unauthorized Control Sphere.

Critical Impact

Shell history containing credentials, tokens, and secrets is transmitted across trust boundaries to external LLM providers without redaction.

Affected Products

  • Goose with terminal integration feature enabled
  • Deployments configured with external LLM provider endpoints
  • Environments where developers or administrators use the terminal to handle secrets

Discovery Timeline

  • 2026-10-22 - CVE CVE-2026-111079 published to NVD
  • 2026-10-22 - Last updated in NVD database

Technical Details for CVE-2026-111079

Vulnerability Analysis

Goose integrates with user terminals to provide context-aware assistance from an LLM. When terminal integration is active, the application reads shell command history and includes it in query payloads sent to the configured LLM provider. The data leaves the local trust boundary unmodified.

Shell history commonly contains high-value secrets. Developers paste API tokens into curl commands, export environment variables inline, or run database clients with passwords on the command line. Each of these entries becomes part of the context stream Goose sends to the external provider.

The exposure extends beyond the immediate transmission. LLM providers typically log inbound prompts for abuse monitoring, model improvement, or billing. A single query can therefore place long-lived credentials into third-party log retention systems the user does not control.

Root Cause

The root cause is the absence of a sanitization layer between the shell history collector and the outbound LLM request builder. The application trusts captured history as safe context and performs no pattern matching, entropy analysis, or deny-list filtering against known secret formats before transmission.

Attack Vector

Exploitation requires user interaction, specifically the act of running commands containing secrets while terminal integration is enabled. A malicious or compromised LLM provider, a man-in-the-middle on the request path, or any party with access to the provider's logs can harvest the leaked data. The scope is changed because secrets collected on one system can authenticate attackers to other services and accounts.

The vulnerability manifests in the terminal history capture and transmission path. See the Red Hat CVE-2026-111079 Advisory and Red Hat Bugzilla Report #2554025 for technical details.

Detection Methods for CVE-2026-111079

Indicators of Compromise

  • Outbound network connections from Goose processes to LLM provider endpoints containing payload sizes consistent with shell history dumps.
  • Secrets appearing in LLM provider dashboards, usage logs, or billing records that match credentials used on hosts running Goose.
  • Unexpected authentication events using credentials that were recently typed into a terminal integrated with Goose.

Detection Strategies

  • Inspect egress traffic from workstations running Goose for request bodies that include shell command fragments or environment variable syntax.
  • Audit LLM provider accounts for prompt logs containing high-entropy strings, token prefixes such as ghp_, sk-, or AKIA, and password-like patterns.
  • Correlate credential rotation events with periods of Goose terminal integration usage to identify likely leaks.

Monitoring Recommendations

  • Enable data loss prevention (DLP) rules that flag outbound HTTPS requests containing known secret patterns from developer endpoints.
  • Monitor process telemetry for Goose reading shell history files such as .bash_history, .zsh_history, or .psql_history.
  • Track creation and use of API tokens tied to users who run Goose with terminal integration enabled.

How to Mitigate CVE-2026-111079

Immediate Actions Required

  • Disable the Goose terminal integration feature until a sanitizing patch is applied.
  • Rotate all credentials, API tokens, and secrets that may have been typed into terminals while the feature was active.
  • Review LLM provider logs for exposed secrets and request deletion where supported by the provider.
  • Restrict Goose usage to non-production accounts and non-sensitive environments during remediation.

Patch Information

Refer to the Red Hat CVE-2026-111079 Advisory for the current patch status and updated package availability. Apply vendor-supplied updates that introduce sanitization of shell history before transmission to LLM providers.

Workarounds

  • Turn off terminal integration in the Goose configuration until a fix is available.
  • Use dedicated secret managers and avoid typing credentials directly into terminals integrated with LLM tooling.
  • Configure shell settings such as HISTIGNORE and leading-space ignores to keep sensitive commands out of history files.
  • Route Goose traffic through an egress proxy that redacts known secret patterns from outbound request bodies.
bash
# Configuration example: prevent sensitive commands from entering shell history
export HISTCONTROL=ignorespace:ignoredups
export HISTIGNORE='*AWS_*:*TOKEN*:*PASSWORD*:*SECRET*:*KEY*'

# Disable Goose terminal integration (consult vendor documentation for exact key)
# Example illustrative setting:
#   terminal_integration.enabled = false

Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

Sources: National Vulnerability Database, CISA KEV, Exploit-DB

Experience the Most Advanced Cybersecurity Platform

See how the world’s most intelligent, autonomous cybersecurity platform can protect your organization today and into the future.