The Agent Skills Directory

[COMMAND_EXECUTION]: The skill dispatches an implementer sub-agent instructed to write and execute tests (using TDD principles) to verify code functionality. This involves the execution of generated code within the agent's environment.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by consuming task requirements from external markdown files (docs/plans/*.md). Malicious content within these plans could attempt to influence the sub-agent's behavior during code generation or command execution.
Ingestion points: Files located at docs/plans/{Story-Key}/analysis.md and plan.md are used as the primary source of task definitions.
Boundary markers: The prompts do not specify the use of delimiters or specific instructions to ignore embedded commands within the task text when passing it to the implementer sub-agent.
Capability inventory: The workflow includes filesystem write access, Git operations, Jira issue retrieval (jira_get_issue), and the ability to execute shell commands for testing purposes.
Sanitization: The skill mitigates risks through a structured 'Spec Reviewer' and 'Code Quality Reviewer' loop, where separate sub-agents verify the implementation against the original requirements and check for quality/security issues before completion.
[DATA_EXPOSURE]: The skill uses the jira_get_issue tool to interact with external Jira instances to fetch issue details and handle task blockers.

aile-subagent-dev