The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The .env file contains a hardcoded administrative access token (MIXIMODEL_ADMIN_TOKEN) in plaintext. This token provides full administrative control over the associated blog API.
Evidence: .env contains MIXIMODEL_ADMIN_TOKEN=oat_MzY5.aC1KQnFMRHB6dUh3Z21PMjBXVWN3d0tHTERpaTNSY3I1RklPel9MSzk2MDI4OTIwMw.
[DATA_EXFILTRATION]: The skill includes an api subcommand that enables the agent or user to perform arbitrary HTTP requests (GET, POST, PUT, DELETE) to any path on the configured remote server using the administrative token.
Evidence: scripts/blog.py implementation of cmd_api using _json_request.
[DATA_EXFILTRATION]: The skill is designed to read local files (Markdown and images) and transmit them to a remote server, which represents a potential data exit path if configured with a malicious URL.
Evidence: scripts/blog.py reading files and using urllib.request.urlopen to transmit data.
[PROMPT_INJECTION]: The skill ingests and processes untrusted local data from Markdown files without sufficient sanitization or boundary markers, creating a surface for indirect prompt injection.
Ingestion points: scripts/blog.py reads full file content in _parse_article_file.
Boundary markers: None identified; content is parsed directly for metadata and body.
Capability inventory: The skill has network access (urllib.request) and administrative API access.
Sanitization: No sanitization is performed on the Markdown content before it is processed by the logic.

miximodel