accint-frames

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill requires the agent to read and embed a submit_token (a credential-like value) verbatim into acc_act input payloads, which forces the LLM to handle/output secret values directly.

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s required runtime workflow (“List the queue: acc frames” then for each open/waiting frame read its typed hole + retrieved context and submit proposal_text) indicates the LLM ingests previously stored frame content and retrieved context that are not authored by the operating user, i.e., outsider-authored free text coming from the deliberation queue/frame records at runtime.