submit-to-inbox

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill explicitly instructs using a GitHub PAT in API mode (PUT to Contents API with an Authorization: Bearer header) which requires the agent to read and embed the secret token into requests/commands, creating a high risk of verbatim secret exposure.

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

• Third-party content exposure detected (high risk: 0.75). RUNTIME path: the skill may read outsider-authored free text from user-provided “素材(链接/文字/图片)” and/or fetched public web pages via /browse/WebFetch, then includes that extracted prose in the generated Markdown frontmatter/body that is sent to the LLM context for submission.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)