Skip to content

Security: agentstation/ago

Security

SECURITY.md

Security Policy

Supported versions

The latest tagged release receives security fixes. Development snapshots and older releases do not receive separate security support.

Reporting a vulnerability

Email security@agentstation.ai. Do not open a public issue for a suspected vulnerability.

Include the affected version (ago -version), the Go toolchain version, the command you ran, a minimal reproduction, and the security effect. Remove any proprietary source from your report — a synthetic reproduction is always preferable.

We will confirm receipt, assess the report, and coordinate remediation and disclosure with you. Please do not disclose an unresolved report publicly before that coordination is complete.

Threat model

ago is a static analyzer. It reads Go source and configuration and writes a report. It executes no code from the packages it analyzes.

In scope:

  • A crafted .ago.yml or //ago:ignore directive that causes ago to crash, hang, or consume unbounded memory.
  • A crafted source file that causes ago to silently skip analysis, so a violation goes unreported while the exit status stays 0.
  • Output injection — a finding message, file path, or rule name that escapes its encoding in -format sarif or -format github and forges an annotation or a code-scanning alert.
  • Path traversal through an exclude pattern or a package pattern.

Out of scope:

  • ago invokes the Go toolchain through golang.org/x/tools/go/packages, which loads and builds package metadata. Running ago on untrusted source is equivalent to running go list on it, and carries the same risk. Do not run ago on source you would not run go build on.
  • Vulnerabilities in the Go toolchain or in golang.org/x/tools. Report those upstream; we will pick up the fix on the next release.
  • A rule producing a false positive or a false negative. That is a correctness bug — please file it as a normal issue.

There aren't any published security advisories