The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on the Bash tool to execute the included Python script scripts/medium.py. This is standard for local script execution and is implemented without unsafe shell patterns.
[EXTERNAL_DOWNLOADS]: The script scripts/medium.py fetches image data from external URLs during the publishing process. To mitigate risk, it includes an SSRF guard in the _assert_public_url function that resolves hostnames and verifies they do not point to private, loopback, or non-public IP ranges before initiating a request.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data retrieved from the Medium API and local user-provided files, representing an attack surface for indirect prompt injection where instructions embedded in post content could influence the agent.
Ingestion points: External content enters via cmd_articles (from the Medium API) and cmd_publish (from local Markdown files) in scripts/medium.py.
Boundary markers: There are no explicit delimiters or specific instructions for the agent to disregard natural language commands within the processed data.
Capability inventory: The skill has access to the Bash tool and file system through its standard operation.
Sanitization: The script converts Markdown content into Medium's structured paragraph delta format, which provides structural isolation but does not perform semantic filtering of the text content.

medium