The Agent Skills Directory

[SAFE]: The skill implements a controlled publishing process that requires pull requests and prohibits direct pushes to the main branch, maintaining repository integrity.
[COMMAND_EXECUTION]: The skill uses the GitHub CLI (gh) and specific MCP tools to perform Git operations and create pull requests, which are standard and appropriate for the intended workflow.
[INDIRECT_PROMPT_INJECTION]: The skill processes article content from local files, which presents a surface for indirect instructions. This risk is minimized by required frontmatter validation and mandatory review gate checks prior to publication.
Ingestion points: article content at content/posts/<slug>.mdx (SKILL.md)
Boundary markers: Frontmatter schema validation is required, though explicit delimiters for raw content are not defined.
Capability inventory: Git operations (branch, commit) and Pull Request creation via gh pr create or publish_post MCP.
Sanitization: Requires execution of strand validate and verification of review logs before proceeding with the workflow.

strand-publish