The Agent Skills Directory

[DATA_EXFILTRATION]: The skill accesses the sensitive environment variable $LABNOCTURNE_API_KEY. This credential is transmitted to an external host (images.labnocturne.com) via an Authorization header. Furthermore, the target host is determined by the $LABNOCTURNE_BASE_URL variable; if this environment variable is compromised, the API key could be exfiltrated to an unauthorized server.
[EXTERNAL_DOWNLOADS]: If the required API key is not present locally, the skill automatically performs a network request to https://images.labnocturne.com/key to fetch a temporary credential.
[COMMAND_EXECUTION]: The skill utilizes the curl system utility to perform network operations. It constructs command strings by dynamically interpolating user-controlled parameters (limit, offset, sort) into the URL, which could lead to command manipulation if the agent does not properly sanitize these inputs.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through data ingestion.
Ingestion points: External data enters the agent's context through the JSON response body of the image listing API (processed in SKILL.md).
Boundary markers: The skill does not implement boundary markers or instructions to treat the remote metadata (filenames, IDs) as untrusted content.
Capability inventory: The skill possesses the capability to execute system commands via curl and perform further network requests.
Sanitization: There is no evidence of sanitization or validation performed on the filenames or other metadata retrieved from the remote service before they are displayed to the user or used in subsequent logic.

image-files