terse-improve

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

• Third-party content exposure detected (medium risk: 0.30). The required workflow described in SKILL.md ingests only production event payloads when running terse history/terse replay/terse test run (e.g., trigger payloads and full chat history), and those payloads are attacker-controlled via the monitored trigger source rather than first-party trusted documentation.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill explicitly instructs the agent at runtime to fetch https://docs.useterse.ai/llms.txt to discover and then pull live docs pages which would be ingested and used to shape prompts/instructions, so external content can directly control agent behavior.