GitHub code scanning for AI applications
A GitHub code scanning detector for prompt injection, sensitive data disclosure and excessive agency in AI apps, run on pull requests.
Overview
GitHub released code scanning for AI applications in public preview on October 28, 2026. It scans AI apps and agents for indirect prompt injection (malicious instructions in files, web pages or issue comments an agent reads), sensitive information disclosure through missing authorization controls, and excessive agency where an agent has more permissions or autonomy than its task needs. Scans run automatically on pull requests alongside CodeQL and show findings in the PR with mitigation guidance. It requires a GitHub Advanced Security license, repositories already using code scanning, and is not available on GitHub Enterprise Server.
This is a sourced directory entry. Features and capabilities are reported in the linked sources; this entry does not document independent Aymorphic testing.
Key features
- Detects indirect prompt injection, sensitive information disclosure and excessive agency
- Runs automatically on pull requests next to CodeQL
- Inline mitigation guidance in the PR
- Requires GitHub Advanced Security; not on Enterprise Server
Use cases
- Reviewing agent code for over-broad tool permissions before merge
- Catching prompt-injection paths in LLM apps that read untrusted content
Pricing
Included with a GitHub Advanced Security license during public preview; no separate price listed.
Pricing reflects the source record and may change. Confirm current terms with the provider.
Dates and sources
- Released
- 28 Oct 2026
- Record updated
- 29 Oct 2026
- Record last checked
- 29 Oct 2026