The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input such as file paths and message text, creating a surface for potential indirect prompt injection attacks.
Ingestion points: The sendImage method in scripts/feishu-image.js accepts imagePath, receiveId, and text parameters from the agent context.
Boundary markers: There are no explicit delimiters or safety instructions present to differentiate between trusted agent instructions and potentially malicious content within the processed data.
Capability inventory: The skill script scripts/feishu-image.js performs local file reads via fs.readFileSync and makes network requests via https.request.
Sanitization: The skill does not validate or sanitize the imagePath input to ensure it targets intended files or to prevent path traversal before reading from the filesystem.
[DATA_EXFILTRATION]: The skill reads local file content and transmits it to an external messaging service.
Evidence: The script scripts/feishu-image.js reads file data from the local disk and uploads it to the Feishu API at https://open.feishu.cn. This domain belongs to a well-known enterprise messaging platform and represents the primary intended functionality of the skill.

feishu-image