# CVE-2026-111079: Three Agent Advisories: Goose Sent Raw Terminal History to the LLM Provider (CVE-2026-111079), a Colon in a Hermes Agent Webhook Route Borrowed Another Route's Tools (CVE-2026-114463), and OpenClaw CDP Discovery Reached Blocked Hosts (CVE-2026-62197)

> CVE-2026-111079 is a high-severity authentication bypass issue (CVSS 7.4) in Goose with the terminal integration enabled, an AI agent. Affected: Goose with the terminal integration enabled (no fixed version named); Hermes Agent webhook gateway before v0.22.2; OpenClaw before 2026.6.6.. Fixed in 0.22.2.

- **Canonical URL:** https://catalog.anomity.ai/vuln/cve-2026-111079
- **Aliases:** ti-153, GHSA-3x84-qq85-fj65
- **Severity:** high
- **CVSS:** 7.4
- **Attack type:** Authentication bypass, Server-side request forgery, Information disclosure
- **Component:** Goose with the terminal integration enabled (AI agent)
- **Catalog entry:** https://catalog.anomity.ai/agent/hermes-agent
- **Exploited in the wild:** no known exploitation
- **Published:** 2026-10-22
- **Last verified:** 2026-11-01
- **JSON:** https://catalog.anomity.ai/api/entries/vuln/cve-2026-111079

## Summary

Red Hat published CVE-2026-111079 (CVSS 7.4) on 22 October 2026 for Goose: when the terminal integration is enabled, Goose captures shell command history and sends it verbatim, unredacted, to the configured LLM provider with each query, so tokens, passwords or secrets typed or exported in that terminal leave the machine and can end up in provider logs. Red Hat's record names no fixed version. CVE-2026-114463, published 21 October for Nous Research Hermes Agent, is in the webhook gateway: it builds a session chat_id as webhook:{route_name}:{delivery_id} and later re-derives the authorizing route by splitting that string, so a route whose name contains a colon is authorized as a different, higher-privileged prefix route, and an attacker holding the first route's HMAC secret gets the other route's toolsets. The fix is referenced in release v0.22.2. OpenClaw GHSA-3x84-qq85-fj65 (CVE-2026-62197, published 21 October; GitHub rates it medium, CVSS 8.5) let browser CDP discovery accept WebSocket URLs that OpenClaw policy should have blocked, so a lower-trust caller could reach blocked network destinations; it is fixed in 2026.6.6.

## Affected versions

| Affected | Fixed in |
| --- | --- |
| Goose with the terminal integration enabled (no fixed version named); Hermes Agent webhook gateway before v0.22.2; OpenClaw before 2026.6.6. | 0.22.2 |

## Am I affected by CVE-2026-111079?

You are affected if you run Goose with the terminal integration enabled (no fixed version named); Hermes Agent webhook gateway before v0.22.2; OpenClaw before 2026.6.6.. Versions 0.22.2 and later include the fix.

## How do I fix CVE-2026-111079?

In Goose, turn the terminal integration off on machines where secrets pass through the shell, keep secrets out of command lines (read them from files or a secret manager rather than exporting or pasting them), and rotate any credential typed in a terminal while the integration was on. Upgrade Hermes Agent to v0.22.2 or later, avoid colons in webhook route names, and give each route its own HMAC secret and the narrowest toolset it needs. Upgrade OpenClaw to 2026.6.6 or later and keep browser and CDP features restricted to trusted operators.

## How do I find every machine running Goose with the terminal integration enabled?

Anomity inventories the AI agents, MCP servers, IDE extensions and skills on each developer device, so you can see which machines run Goose with the terminal integration enabled and which version, then enforce a policy on it.

## References

- article: https://access.redhat.com/security/cve/CVE-2026-111079
- cve: https://nvd.nist.gov/vuln/detail/CVE-2026-111079
- advisory: https://github.com/advisories/GHSA-3x84-qq85-fj65

## Related vulnerabilities

- [GHSA-53mr-w429-c597](https://catalog.anomity.ai/vuln/ghsa-53mr-w429-c597) — high
- [CVE-2026-103091](https://catalog.anomity.ai/vuln/cve-2026-103091) — critical
- [CVE-2026-109527](https://catalog.anomity.ai/vuln/cve-2026-109527) — high
- [CVE-2026-82533](https://catalog.anomity.ai/vuln/cve-2026-82533) — critical
- [CVE-2026-19516](https://catalog.anomity.ai/vuln/cve-2026-19516) — critical
- [CVE-2026-64849](https://catalog.anomity.ai/vuln/cve-2026-64849) — critical

---

Source: Anomity Security Database (https://catalog.anomity.ai/vuln). Detect affected AI tools on developer devices: https://anomity.ai/?utm_source=catalog&utm_medium=vuln-md&utm_campaign=cve-2026-111079