The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data, such as meeting transcriptions and articles, which is a common vector for indirect prompt injection attacks.
Ingestion points: SKILL.md directs the agent to read from user-provided text, output/transcription_plain.txt, and external web links.
Boundary markers: The instructions lack specific delimiters or warnings to the model to ignore potential instructions embedded within the processed content.
Capability inventory: The skill possesses file-read and file-write capabilities, along with the ability to execute a local Python script.
Sanitization: No sanitization logic is implemented to filter or escape the content being summarized before it is processed by the agent.
[COMMAND_EXECUTION]: The skill requires the execution of a bundled Python script to perform document conversion, which involves the agent running shell commands.
Evidence: SKILL.md explicitly instructs the agent to run python "<skill_dir>/scripts/md_to_word.py" "<path_to_summary.md>" for Word export functionality.

hebrew-tech-lecture-summary