The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates the download and installation of external software from the author's NPM (@mvanhorn/printing-press-library) and GitHub (github.com/mvanhorn/printing-press-library) repositories.\n- [REMOTE_CODE_EXECUTION]: The skill encourages the use of npx and go install to fetch and execute code from remote sources. While these are vendor-owned resources, they involve executing remote content at runtime.\n- [DATA_EXFILTRATION]: The CLI tool includes a --deliver webhook:<url> feature that allows the output of any command to be transmitted to an arbitrary external URL. This provides a direct mechanism for data exfiltration if the agent is manipulated into using a malicious destination.\n- [COMMAND_EXECUTION]: The skill includes several features that could lead to unauthorized command or file access:\n
The --deliver file:<path> flag enables writing data to arbitrary file paths, which could be used to overwrite configuration files or establish persistence.\n
The skill takes $ARGUMENTS and interpolates them directly into a shell command (instagram-pp-cli <command> [subcommand] [args] --agent), creating a potential surface for command injection if the agent does not sanitize input.\n- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it retrieves and processes untrusted data from the Instagram API.\n
Ingestion points: The comments list, media list, and accounts get commands ingest data authored by third-party Instagram users (e.g., comments, bios, and captions).\n
Boundary markers: The CLI wraps output in a JSON envelope, but there are no instructions or delimiters to prevent the agent from following commands embedded in the retrieved data.\n
Capability inventory: The skill has access to the Bash and Read tools, which could be abused if the agent is manipulated by malicious data.\n
Sanitization: No evidence of data sanitization or filtering is present for the ingested Instagram content.

pp-instagram