The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources which could potentially influence agent behavior.
Ingestion points: Reads GitHub issue bodies via gh issue view and Slack thread replies via slack_read_thread to resolve stakeholders and check release progress.
Boundary markers: Uses specific keyword patterns (e.g., "I can handle [product]") to identify relevant information, but does not implement robust delimiters to separate external content from instructions.
Capability inventory: Includes the ability to execute gh and git commands, write local state files, and post to Slack channels.
Sanitization: There is no evidence of explicit escaping or validation of the content parsed from external GitHub issues or Slack messages before it is processed or stored in the state file.
[COMMAND_EXECUTION]: The skill makes extensive use of the GitHub CLI (gh) and git to automate repository management, issue editing, and pull request tracking within the elastic organization.
[DATA_EXPOSURE]: The skill maintains session state in a local file at ~/.elastic-docs/stack-release-state.json. While this file contains information about active releases and PR statuses, it is scoped to the user's home directory and the vendor's specific toolset.

docs-stack-release