The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data from GitHub, including pull request titles, issue descriptions, and discussion comments, which are incorporated into the generated standup drafts. This creates a surface for indirect prompt injection where malicious instructions embedded in these external sources could attempt to influence the agent's behavior during the standup generation process.
Ingestion points: Activity is fetched from GitHub repositories using the gh CLI within the get-standup-context.sh script.
Capability inventory: The skill can execute shell scripts, write to the local file system, and post comments to GitHub Discussions.
Boundary markers: The skill uses structured markdown (links and bullets) but lacks explicit instructions to sanitize or ignore embedded commands within the fetched content.
Sanitization: There is no evidence of filtering external strings before they are interpolated into the prompt for summarization.
[COMMAND_EXECUTION]: The skill executes bundled shell scripts (get-standup-context.sh and post-standup-comment.sh) to interact with the GitHub API. These scripts utilize jq to safely handle JSON data and pass parameters securely to the gh CLI.
[SAFE]: The skill implements security best practices by creating a private cache directory for standup drafts and explicitly setting restrictive permissions (chmod 700). This ensures that sensitive information, such as session summaries and GitHub activity, is not accessible to other users on a shared system.
[SAFE]: All network operations are directed at GitHub, which is a well-known service, for fetching developer activity and posting standup reports via official API endpoints.

generate-standup