https://vulnerability.circl.lu/sightings/feed 2026-09-13T15:23:40.215192+00:00 Vulnerability-Lookup info@circl.lu python-feedgen Contains only the most 10 recent sightings. https://vulnerability.circl.lu/sighting/7ce09b08-f00f-46a9-90c9-f899b5ae0cdc/export 2026-09-13T15:23:40.240655+00:00 Automation user https://vulnerability.circl.lu/user/automation {"uuid": "7ce09b08-f00f-46a9-90c9-f899b5ae0cdc", "vulnerability_lookup_origin": "1a89b78e-f703-45f3-bb86-59eb712668bd", "author": "9f56dd64-161d-43a6-b9c3-555944290a09", "vulnerability": "CVE-2026-62676", "type": "seen", "source": "https://gist.github.com/alon710/43bf5918e3d482ece766ad1a3e45b23b", "content": "# CVE-2026-62676: CVE-2026-62676: Security Guardrail Policy Bypass via Shell-Command Parser Flaws in Omnigent AI Agent Framework\n\n> **CVSS Score:** 7.1\n> **Published:** 2026-09-02\n> **Full Report:** https://cvereports.com/reports/CVE-2026-62676\n\n## Summary\nA high-severity security guardrail policy bypass vulnerability was identified in the Omnigent AI agent framework. The shell-command parser failed to correctly identify gated commands wrapped in certain command modifiers, combined interpreter flags, or command substitutions. Consequently, the default fail-open behavior allowed arbitrary execution of restricted operations.\n\n## TL;DR\nFlaws in Omnigent's shell-command parser allow command wrappers, interactive flags, or command substitutions to bypass security guardrails and run forbidden operations due to a fail-open default.\n\n## Exploit Status: POC\n\n## Technical Details\n\n- **CWE ID**: CWE-184\n- **Attack Vector**: Network\n- **CVSS**: 7.1\n- **EPSS Score**: 0.00295\n- **EPSS Percentile**: 21.63%\n- **Impact**: Guardrail Policy Bypass\n- **Exploit Status**: POC\n- **KEV Status**: Not Listed\n\n## Affected Systems\n\n- Omnigent Framework\n- **Omnigent**: < v0.3.0 (Fixed in: `v0.3.0`)\n\n## Mitigation\n\n- Upgrade the Omnigent AI framework to version v0.3.0 or later.\n- Restrict shell interpreters available within the execution environment.\n- Configure branch protection rules on repositories mapped to coding agents.\n\n**Remediation Steps:**\n1. Identify all deployment instances running Omnigent framework versions prior to v0.3.0.\n2. Update dependencies in requirements.txt or poetry configs to target omnigent>=0.3.0.\n3. Apply the patch manually to policies/builtins/_shell.py if direct framework upgrades are barred.\n4. Verify parsing improvements using test suites executing nested command configurations.\n\n## References\n\n- [GitHub Security Advisory GHSA-7mqg-cx4g-x2rf](https://github.com/omnigent-ai/omnigent/security/advisories/GHSA-7mqg-cx4g-x2rf)\n- [NVD Entry for CVE-2026-62676](https://nvd.nist.gov/vuln/detail/CVE-2026-62676)\n- [Vulnerability Fix Pull Request](https://github.com/omnigent-ai/omnigent/pull/389)\n- [Vulnerability Fix Commit](https://github.com/omnigent-ai/omnigent/commit/1a05b7b139ef504bf2be89bf37918abe104fb95c)\n- [Omnigent v0.3.0 Release](https://github.com/omnigent-ai/omnigent/releases/tag/v0.3.0)\n\n\n---\n*Generated by [CVEReports](https://cvereports.com/reports/CVE-2026-62676) - Automated Vulnerability Intelligence*", "creation_timestamp": "2026-09-03T09:31:44.273501Z"} 2026-09-03T09:31:44.273501+00:00