The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from project files which could theoretically contain instructions to influence the agent's output.
Ingestion points: The agent reads project documentation, briefs, and task logs (ticket.md, journal.md, etc.) from the .task/ directory.
Boundary markers: The skill does not define specific delimiters or "ignore instructions" prompts for these source files.
Capability inventory: The skill's capabilities are limited to reading local files and writing a Markdown report to the designated task folder. It does not have network access, shell execution, or privilege escalation capabilities.
Sanitization: Data is processed for text-based analysis and stored in a documentation file; no executable logic is derived from these inputs.
[SAFE]: All referenced resources, such as the retro skill and specific flow- components, are internal to the skill environment or part of the user's project structure. No external network operations or remote code downloads are present.

flow-retro