The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/profile.py utilizes the subprocess module to execute system commands including python, node, psql, and mysql. It accepts user-provided file paths as arguments and executes them directly using these runtimes, which allows for the execution of arbitrary local code files.
[DATA_EXFILTRATION]: The skill's database analysis components in scripts/db-slow-query.sql and scripts/profile.py query internal system views (pg_stat_statements for PostgreSQL and performance_schema for MySQL) to retrieve the full text of queries. This can lead to the exposure of sensitive application data, PII, or security tokens if the applications being analyzed do not use parameterized queries.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes and displays raw SQL query text from external database logs which could contain malicious instructions.
Ingestion points: Raw SQL query strings are ingested from database performance tables into the agent's context via scripts/profile.py and scripts/db-slow-query.sql.
Boundary markers: No delimiters or instructions are present to ensure the agent ignores embedded commands within the log data.
Capability inventory: The skill possesses command execution capabilities through subprocess calls in scripts/profile.py.
Sanitization: No filtering, escaping, or validation is performed on the SQL data before it is presented to the agent.

performance-optimizer