The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes local shell commands such as git diff, ls, and make eval-all to extract version metrics and repository status. These commands are part of the documented execution flow to provide data for the promotion posts.
[DYNAMIC_EXECUTION]: The skill includes a Python code snippet designed to verify the visual width of ASCII table lines. This script uses the unicodedata module to ensure that Unicode box-drawing characters are correctly aligned within a 72-character limit.
[INDIRECT_PROMPT_INJECTION]: The skill reads from local files like CHANGELOG.md and CLAUDE.md to populate release information. While these are local files, they represent an ingestion surface for potentially untrusted data.
Ingestion points: CHANGELOG.md, CLAUDE.md, and CLI tool outputs in SKILL.md.
Boundary markers: No specific delimiters or warnings are used when interpolating this data into the generated post templates.
Capability inventory: The skill can write files to the scratchpad/ directory, execute shell commands, and run Python code (found in references/templates.md).
Sanitization: No explicit sanitization or filtering of the ingested text is performed.

promote