The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/mineru_fulltext_extract.py accepts a --file argument and reads the content of local files. While restricted by an extension allowlist, it enables the reading and processing of arbitrary local documents.
[COMMAND_EXECUTION]: The script includes an --insecure flag that disables TLS certificate verification by using ssl._create_unverified_context(). This bypasses standard network security protocols and makes the connection vulnerable to man-in-the-middle (MitM) attacks.
[DATA_EXFILTRATION]: The skill transmits the contents of local documents to an external API endpoint defined by the UNSTRUCTURED_API_BASE_URL environment variable, which allows potentially sensitive data to be sent to a remote server.
[PROMPT_INJECTION]: The skill processes untrusted document content and returns it as plain text to the agent, creating a surface for indirect prompt injection. \n
Ingestion points: Document content is read from local files in scripts/mineru_fulltext_extract.py. \n
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are used when processing the document text. \n
Capability inventory: The script has the capability to read local files and perform network POST requests. \n
Sanitization: The extracted text is returned without sanitization or filtering.

document-granular-decompose