Gitingest: Arbitrary file read via symlink in blob ingestion

Summary

Gitingest through version 0.3.1 follows symlinks during single-file blob ingestion. An unauthenticated remote attacker can commit a symlink to a public repository and submit its blob URL, causing the service to return any file the service user can read, up to 10 MB. The CVSS 4.0 score is 8.7 (High), mainly for confidentiality impact.

Affected

• Gitingest up to and including 0.3.1
• Vulnerable component: ingest_query, single-file blob ingestion

Fixed in

Unknown.

Workaround

• Do not expose Gitingest to the internet or to untrusted users; place it behind VPN, SSO or an authenticating reverse proxy
• Run the service as a dedicated low-privilege user without access to secrets, .env files, SSH keys or other repositories
• Keep secrets out of directories the service can read
• Restrict or disable single-file blob ingestion if the deployment allows it

Exploitation

• CISA SSVC: exploitation status PoC, automatable: yes, technical impact: partial
• A third-party write-up exists; its content has not been reviewed
• No evidence of exploitation in the wild found; not in CISA KEV
• EPSS about 0.4%

Our recommendation

• Inventory all Gitingest installations, including pip packages and containers
• Check whether any instance is reachable from the internet and remove public exposure immediately
• Limit access to authenticated internal users only
• Ensure the service account cannot read secrets or sensitive files
• Monitor the upstream repository and the VulnCheck advisory for a fixed release and apply it as soon as available

Public exploits iAttack code is publicly available (PoC on GitHub, Metasploit or Exploit-DB).

PoC-in-GitHub, Metasploit and Exploit-DB list nothing for this CVE.

Research sources

• VulnCheck advisory www.vulncheck.com
• Gitingest GitHub repository github.com
• Vulnerable source line (ingestion.py) github.com
• Third-party write-up (HackMD) hackmd.io
• VulDB entry vuldb.com
• Vulners CVE record vulners.com

Researched and written up in English and German by an AI (Claude by Anthropic) based on NVD, CISA, EPSS and the sources listed. The vendor information is authoritative. Researched on 1 Nov 2026, 02:44.