The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from implementation tickets and shared policy files.
Ingestion points: Accesses and interprets content from workflow/RULES.md, workflow/SPECIALISTS.md, workflow/TASK-FOLDER.md, and external ticket descriptions.
Boundary markers: The skill relies on procedural checkpoints but does not utilize explicit delimiters or 'ignore-previous-instructions' wrappers for ticket content.
Capability inventory: The agent can execute Git commands, run test suites, and invoke subagents to perform code reviews or implementation tasks.
Sanitization: No explicit sanitization or filtering of the ticket data is described in the workflow.
[COMMAND_EXECUTION]: The skill uses shell commands and local scripts as part of its standard operating procedure.
Evidence: Executes git status and git diff to maintain state records in TASK-FOLDER.md.
Evidence: Runs a local Python script workflow/scripts/compare_failures.py to evaluate test output differences.

workflow-execute