deploy-hermes-slack-agent

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

• Third-party content exposure detected (high risk: 0.85). Outsider free text can enter the LLM context via Slack Events/Interactivity at runtime: the user’s Slack message content (authored by non-operating-user parties) is ingested by the Hermes Slack agent and then provided to the agent’s LLM as the prompt.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly requires running runtime npx commands (npx @truefoundry/tfy-hermes-agent validate|compile|deploy ...), which fetch and execute code from the npm registry during validation/compilation/deploy and are required for the workflow, so this is a runtime external dependency that executes remote code.