The Agent Skills Directory

[PROMPT_INJECTION]: The skill includes scripts for parsing and analyzing logs and traces (scripts/log-aggregator.py and scripts/trace-analyzer.sh) which process data from external sources.
Ingestion points: Files provided as command-line arguments to the aggregation and analysis scripts.
Boundary markers: No specific delimiters or instructions are provided to the agent to treat the parsed content as potentially untrusted or to ignore instructions embedded within the logs.
Capability inventory: The skill possesses capabilities for file reading, system metric collection via psutil, and shell command execution.
Sanitization: The scripts use standard JSON parsing and regular expressions to extract data fields but do not perform sanitization on the message content for downstream consumption by an LLM.
[COMMAND_EXECUTION]: The scripts/trace-analyzer.sh script executes shell commands and the jq utility on user-provided file paths. While arguments are handled with quoting to prevent simple injection, it represents an execution surface for analyzing local files.

monitoring-observability