The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a Python script scripts/ingest_markdown_doc.py that processes user-controlled arguments, including file paths.
[DATA_EXFILTRATION]: The scripts/ingest_markdown_doc.py script reads the content of a file specified by the --source argument and writes it to a new location in the docs/ folder. Since there is no validation to ensure the source path is within a safe directory, an attacker could trick the agent into reading sensitive files (such as .env, SSH keys, or system configurations) and exposing them within the repository's documentation tree.
[INDIRECT_PROMPT_INJECTION]: The skill processes external Markdown files which could contain hidden instructions for the agent.
Ingestion points: The scripts/ingest_markdown_doc.py script reads external content from the file path provided in the --source parameter.
Boundary markers: None. The ingested content is not wrapped in delimiters or accompanied by instructions to the agent to ignore embedded commands.
Capability inventory: The skill has the ability to read arbitrary files (source.read_text) and write to the filesystem (destination.write_text).
Sanitization: None. The script performs no content validation or sanitization before writing the data to the documentation directory.

mintlify-docs-updater