The Agent Skills Directory

[PROMPT_INJECTION]: The skill implements a workflow that ingests data from the local codebase to generate article content, which introduces a surface for indirect prompt injection if the project files contain malicious instructions.
Ingestion points: The skill uses Glob, Grep, and Read tools in Step 2 to orientation itself within the user's repository.
Boundary markers: The workflow utilizes independent subagents to produce structured reports (defined in references/subagent-report-format.md), providing a layer of isolation between raw file content and the final prose synthesis.
Capability inventory: The skill is authorized to use Bash, Write, Edit, and Task tools to manage files and execution.
Sanitization: The publishing script (scripts/publish-blog.sh) uses Python's html.escape to sanitize titles and slugs when updating the blog index.
[COMMAND_EXECUTION]: The skill utilizes local shell scripts for critical workflow gates, including scripts/validate-blog.sh for structural integrity checks and scripts/publish-blog.sh for managing the Git publishing lifecycle (add, commit, push).
[EXTERNAL_DOWNLOADS]: The generated blog post template (assets/blog-template.html) references external ESM libraries for Mermaid diagrams and Shiki syntax highlighting from well-known CDNs (cdn.jsdelivr.net and esm.sh). These resources are fetched by the user's browser at viewing time and do not execute within the agent's environment.

blog-generator