The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the shell command rg --files "$SESSION_ROOT" -g '*.jsonl' to identify input files. The variable $SESSION_ROOT is a user-supplied input path, which is directly interpolated into a shell command for execution.\n- [DYNAMIC_EXECUTION]: The instructions explicitly allow for the creation and use of 'Temporary scripts and intermediate files' during the analysis phase. This implies the agent may generate and execute code (such as Python scripts) at runtime to perform data processing, metric calculations, and topic classifications.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data from session logs which may contain malicious instructions targeting the agent context.\n
Ingestion points: The skill recursively reads all .jsonl session files from a user-specified sessions_root directory.\n
Boundary markers: Absent. The skill does not implement specific delimiters or 'ignore' instructions to isolate message content from the analytical context.\n
Capability inventory: The skill can execute shell commands via rg, write files to the output_path, and dynamically execute generated scripts.\n
Sanitization: While the skill includes privacy-focused masking rules for names, it lacks sanitization mechanisms to filter or escape potentially malicious instruction strings embedded in the message text of the sessions.

ov-session-report