The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources that could contain malicious instructions designed to influence the agent's behavior during summarization.
Ingestion points: The skill ingests data from GitHub release bodies (gh release view), pull request metadata and file lists (gh pr view), and GitHub Project iteration items (gh api graphql).
Boundary markers: The instructions do not define explicit delimiters or instructions for the agent to ignore potentially malicious content embedded in the ingested GitHub data.
Capability inventory: The skill utilizes the gh CLI for repository interaction, curl for network requests to documentation sites, and clipboard tools (pbcopy, xclip, clip.exe) for final output delivery.
Sanitization: There is no evidence of sanitization or filtering of the content pulled from GitHub before the agent processes and summarizes it.
[COMMAND_EXECUTION]: The skill provides the agent with specific shell commands to execute locally.
Evidence: Instructions include the use of gh CLI commands to list and view releases, pull requests, and project board data. It also uses curl to verify documentation link status and system clipboard utilities (pbcopy, xclip, clip.exe) to deliver the final post.
[EXTERNAL_DOWNLOADS]: The skill performs network operations to fetch data and verify resources from external domains.
Evidence: The skill uses gh to fetch release and PR data from github.com and curl to check the availability of documentation pages hosted on mckinsey.github.io (GitHub Pages). These are vendor-owned resources but represent external network activity.

slack-announcement