Microsoft takes the security of our software products and services seriously, which includes all source code repositories managed through our GitHub organizations, which include Microsoft, Azure, DotNet, AspNet and Xamarin.
If you believe you have found a security vulnerability in any Microsoft-owned repository that meets Microsoft's definition of a security vulnerability, please report it to us as described below.
Please do not report security vulnerabilities through public GitHub issues.
Instead, please report them to the Microsoft Security Response Center (MSRC) at https://msrc.microsoft.com/create-report.
If you prefer to submit without logging in, send email to secure@microsoft.com. If possible, encrypt your message with our PGP key; please download it from the Microsoft Security Response Center PGP Key page.
You should receive a response within 24 hours. If for some reason you do not, please follow up via email to ensure we received your original message. Additional information can be found at microsoft.com/msrc.
Please include the requested information listed below (as much as you can provide) to help us better understand the nature and scope of the possible issue:
- Type of issue (e.g. buffer overflow, SQL injection, cross-site scripting, etc.)
- Full paths of source file(s) related to the manifestation of the issue
- The location of the affected source code (tag/branch/commit or direct URL)
- Any special configuration required to reproduce the issue
- Step-by-step instructions to reproduce the issue
- Proof-of-concept or exploit code (if possible)
- Impact of the issue, including how an attacker might exploit the issue
This information will help us triage your report more quickly.
We prefer all communications to be in English.
Microsoft follows the principle of Coordinated Vulnerability Disclosure.
- The agent runtime exposes registered host tools only; native shell, MCP and cross-session memory tools are disabled on both creation and resume. The application container runs as a non-root user.
- ARM writes run only through
ApplyAzureChange, which Agent Framework holds until the user explicitly approves the exact method, URL and body in the application. Azure resource deletion and mutating action POSTs remain blocked. Standard Graph consent tiers are read-only. - Downloads, transcripts, uploads, operation records and outcomes require the caller's application identity. Generated HTML is sandboxed without same-origin access, including direct downloads viewed inline.
- The host does not screen or redact credentials in chat, job, tool or SSE content; recognizing and refusing secrets is left to the model. Do not upload or paste credentials. Existing retained transcripts/uploads are not automatically scrubbed; rotate any disclosed credential and apply the deployment's retention policy.
- Host and browser telemetry are separate pipelines. Agent prompt/response content capture is disabled; browser diagnostic properties and URLs are redacted. Protect the persistent volume and telemetry stores as customer data.
- The current coordination model supports one active app instance. Files on a shared volume do not provide distributed turn gates, approval locking or cooldown coordination.
See reliability contracts for retention periods, verification scope and operational limitations.