The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation specifies a runtime dependency on the agent-reviews package, which is executed via npx. This package is sourced from a third-party GitHub repository (pbakaus/agent-reviews) that is not included in the list of trusted vendors or recognized as an official repository of a well-known service.
[REMOTE_CODE_EXECUTION]: Executing the agent-reviews tool via npx at runtime constitutes remote code execution. The tool is downloaded and executed directly from a remote registry without being vendored, pinned to a specific hash, or audited as part of the skill's static files.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from external sources, including GitHub Issues and Linear issue comments. These sources can be controlled by external actors and may contain malicious instructions intended to manipulate the agent's behavior during the planning process.
Ingestion points: Reads inbound context from GitHub Issues and Linear issues/comments as defined in SKILL.md and references/conventions.md.
Boundary markers: The workflow requires humans to move issues from 'Backlog' to 'Todo', providing a manual review step before implementation starts.
Capability inventory: The skill has access to the gh CLI for repository interaction and Linear MCP tools for creating and updating project specifications.
Sanitization: There are no explicit instructions for sanitizing or filtering input from these external platforms.
[COMMAND_EXECUTION]: The skill utilizes the gh CLI to perform operations such as checking authentication status and viewing repository metadata. While these are functional requirements for the skill's purpose, they represent an active interaction with the user's authenticated git environment.

plan