The Agent Skills Directory

[COMMAND_EXECUTION]: The evaluation script evals/run_real_eval.py uses subprocess.Popen to execute the claude CLI for trigger-rate testing. This is a developer-facing utility for regression testing. The script builds the command as a list to prevent shell injection and implements environment variable sanitization before execution.
[INDIRECT_PROMPT_INJECTION]: The skill processes content from tasks.md and breakdown.md to generate ticket fields, which represents an ingestion surface for potentially untrusted data if the source files are externally controlled.
Ingestion points: Local tasks.md and breakdown.md files read via the Read and Glob tools.
Boundary markers: Absent; instructions do not specify the use of delimiters or 'ignore' warnings for the ingested content.
Capability inventory: The skill can read local files, perform file searches, and hand off data to the bitwarden-atlassian-tools:filing-jira-tickets skill.
Sanitization: No explicit sanitization or validation of the task descriptions or acceptance criteria is described in the prompt logic.

filing-breakdown-tasks