blog-flow

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly pulls prompt and reference files from the public GitHub repo github.com/AgriciDaniel/flow via scripts/sync_flow.py (invoked by /blog flow sync) and required runtime steps state the agent must load references/flow-framework.md and selected prompt files from skills/blog-flow/references/, so untrusted third‑party content is ingested and directly used to choose and drive prompts that can change the agent's actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's sync script (invoked at runtime via /blog flow sync) fetches prompt files from https://github.com/AgriciDaniel/flow which are then written into the skill and directly control the agent's prompts, and the skill requires those synced files to operate.