The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from several sources that could be influenced by external contributors, such as git commit history and community-facing documents.
Ingestion points: Reads files from production/sprints/, design/balance/, and extracts data using git log (references/claude-skill.md).
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat these inputs as untrusted data.
Capability inventory: The skill uses Read, Write, and Bash tools (references/claude-skill.md).
Sanitization: No automated sanitization is performed; the skill relies on a manual review phase (Phase 5) and translation heuristics to filter content.
[COMMAND_EXECUTION]: Executes git log as a fallback mechanism to gather change data between release tags (references/claude-skill.md).

patch-notes