sysdig-remediate

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill explicitly searches and reads code and repository data from GitHub/GitLab (e.g., SKILL.md steps 3/3a/3b and references/source_control.md using gh search, gh api, glab api, reading Dockerfiles/manifests and commit history), thereby ingesting user-generated third‑party content which the agent interprets to decide fixes and create PRs—so untrusted external content can materially influence actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill requires running the external npm package via the command npx -y @sysdig/secure-mcp-server (used to register the Sysdig MCP server), which fetches and executes remote code at runtime and is presented as a mandatory dependency for the skill to operate.