The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs Python dependencies including powermem and python-dotenv from the PyPI registry and utilizes npx to execute external processing tools. These represent standard dependency management from well-known sources.
[COMMAND_EXECUTION]: Multiple shell commands are used to initialize a Python virtual environment, manage local configuration files, and execute scripts for searching and storing writing style profiles.
[DATA_EXFILTRATION]: The skill performs network requests to retrieve content from user-provided URLs. While this is a core feature, it involves outgoing connections to arbitrary third-party domains.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it incorporates untrusted data into its style extraction prompt.
Ingestion points: Content retrieved from external URLs via the baoyu-url-to-markdown tool, as well as text and image inputs.
Boundary markers: The extraction prompt in SKILL.md uses --- delimiters to separate instructions from analyzed content.
Capability inventory: Access to shell command execution for script management, file system writes to the /tmp/ directory, and database persistence via the powermem SDK.
Sanitization: No explicit input validation or filtering is performed on external content before it is interpolated into the prompt.

pmem-style-extraction