The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted external data from Jira tickets and GitHub issues, creating an indirect prompt injection surface where embedded instructions could influence the agent's behavior.
Ingestion points: Data from external platforms enters the workflow in Phase 1 through the fetching-work-item skill and is stored in markdown files within the docs/ directory.
Boundary markers: The skill utilizes structured snapshot sections and heading orders (e.g., ## Metadata, ## Description) defined in its platform-specific playbooks to organize and delimit external content.
Capability inventory: The orchestrator and its subagents have access to file system writes (docs/), codebase inspection (git), local file search (rg), and platform interactions through the gh CLI and Jira MCP tools.
Sanitization: The skill instructions direct the orchestrator to use concise summaries and specific validation checks rather than raw issue content, which serves as a mitigation, although explicit data sanitization or escaping is not implemented.
[COMMAND_EXECUTION]: Subagents are designed to execute standard system utilities including git for repository management, gh (GitHub CLI) for platform interaction, and rg (ripgrep) for code analysis. These operations are consistent with the skill's purpose as a development workflow tool.
[EXTERNAL_DOWNLOADS]: The skill contains references to documentation and installation resources from recognized and trusted organizations such as Anthropic, Atlassian, and GitHub. These links are provided for informational purposes and environment setup.

orchestrating-workflow