The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes Python snippets and shell commands (including rg, sort, and uniq) to aggregate metrics from .claude/session-metrics/metrics.jsonl. It also invokes a Python script from a relative sibling directory: ../session-scan/references/compute-metrics.py.\n- [PROMPT_INJECTION]: The skill identifies ingestion points for potentially untrusted data, qualifying as an indirect prompt injection surface. [Ingestion points]: The skill reads .claude/session-metrics/metrics.jsonl and user-specified files via the --compare flag. [Boundary markers]: Instructions explicitly require the agent to read comparison files separately and evaluate them 'cautiously' against aggregates. [Capability inventory]: Subprocess execution for shell commands and Python-based data processing. [Sanitization]: Content is parsed via standard JSON libraries rather than being executed as instructions.

session-trends