The Agent Skills Directory

Indirect Prompt Injection Surface: The skill ingests untrusted data from external sources such as GitHub issues, pull requests, and source code files from the microsoft/vscode and microsoft/vscode-copilot-chat repositories. While this is necessary for the skill's purpose of documenting features, it presents a surface where instructions embedded in those external materials could influence the agent's behavior.
Ingestion points: The agent reads data from GitHub issues, PRs, and source code using the gh CLI (specified in SKILL.md).
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are specified for the ingested content.
Capability inventory: The skill has the capability to execute gh CLI commands and perform file-write operations within the docs/ folder.
Sanitization: There is no explicit sanitization or filtering mentioned for the external content before it is processed.
Command Execution: The skill utilizes the gh command-line interface to interact with GitHub repositories. This usage is confined to official project repositories and is part of the standard documentation workflow, providing a structured way to gather feature information.
File System Access: The instructions grant the agent permission to create and modify files within the docs/ directory. The skill includes protective guardrails, such as prohibiting direct edits to the generated enterprise/policies.md file, to ensure documentation integrity.

doc-writer