The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the GitHub CLI (gh), date, and git to retrieve work history and repository context. The argument parsing logic includes validation for integers (days) and strictly matches allowed flags, preventing shell injection from user-supplied parameters.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a data aggregator, ingesting external content that could potentially contain malicious instructions.
Ingestion points: The skill reads pull request titles/bodies from GitHub and issue titles from Linear (via MCP).
Boundary markers: Instructions mandate wrapping the final output in four-backtick fences ( ) to isolate the rendered markdown and prevent inner code blocks from breaking the agent's response structure.
Capability inventory: The skill is granted Bash, Read, and Write capabilities. While Write is allowed, the instructions explicitly advise against writing to files unless requested by the user.
Sanitization: The skill implements sanitization rules, such as stripping emojis and trailing punctuation from titles, and a "jargon strip-list" for the general audience mode to ensure the output remains focused on high-level summaries.

changelog