The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The installation section provides the official shell command to download and execute the Zed installation script from https://zed.dev/install.sh. This is a standard and recognized installation method for the software vendor.
[COMMAND_EXECUTION]: The documentation includes instructions for building the editor from source using cargo build, as well as configuration for external formatters like prettier and ruff. These are routine development activities for the target software.
[DATA_EXFILTRATION]: The skill specifically instructs users to configure disabled_globs for sensitive file types such as .env, *.pem, and *.key. This is a defensive security measure designed to prevent the AI assistant from inadvertently accessing credentials or secrets.
[INDIRECT_PROMPT_INJECTION]: The skill describes integrated AI features that can ingest content from the filesystem, terminal output, and editor buffers.
Ingestion points: Users can provide context via slash commands (/file, /tab, /terminal) or by dragging files into the assistant panel.
Boundary markers: The skill does not define specific boundary markers for this ingested data, as it documents the tool's native capabilities.
Capability inventory: The documented environment includes file system access, terminal interaction, and Rust build tools (cargo).
Sanitization: No explicit sanitization of ingested content is described, reflecting the standard operational model of the AI assistant within the editor.

zed