agents-md
Warn
Audited by Snyk on Mar 9, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 0.90). This skill explicitly fetches and ingests the public upstream file at https://raw.githubusercontent.com/robinebers/agents.md/main/AGENTS.md (per SKILL.md install/update steps), and the agent is expected to read and merge that untrusted, user-maintained content into the workspace, which can materially influence merge decisions and file contents.
MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).
- Potentially malicious external URL detected (high risk: 0.90). The skill runtime explicitly downloads and installs upstream agent instructions from https://raw.githubusercontent.com/robinebers/agents.md/main/AGENTS.md using curl, and that fetched AGENTS.md directly defines agent prompts/behavior and is treated as a required upstream dependency.
Audit Metadata