The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from repository files to drive automated modifications, creating a surface for indirect prompt injection.
Ingestion points: The skill reads milestones/{slug}/tasks.md and roadmap.md to extract completion dates, task statuses, and deliverables (documented in SKILL.md Phase 1 & 2).
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands within the processed files are specified.
Capability inventory: The skill performs file creation (_archive/{slug}-summary.md), file modification (roadmap.md), directory deletion (milestones/{slug}/), and repository-wide search-and-replace (documented in SKILL.md Core Objectives).
Sanitization: There is no evidence of sanitization or validation for the strings extracted from tasks.md before they are interpolated into the generated summary or roadmap.
[COMMAND_EXECUTION]: The skill instructs the agent to perform significant file system operations, including directory removal and repository-wide string replacement.
Evidence: The skill requires the removal of the milestones/{slug}/ directory and a global 'grep' and update of all references to the old task paths across the entire repository (SKILL.md Core Objective 4 & 5).

archive-milestone