The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill recommends installing utility packages and referencing code patches from the vendor's official GitHub repository.
Evidence: Instructions in references/automation-and-capture.md to cargo install gpui-storybook-launch and use [patch.crates-io] entries from https://github.com/stayhydated/gpui-storybook/blob/wayland-render-image/Cargo.toml.
[COMMAND_EXECUTION]: The skill directs the user to perform various CLI operations for building, testing, and running the storybook environment.
Evidence: Use of cargo run, cargo install, and just wayland-capture-test commands as described in references/automation-and-capture.md and references/setup-and-configuration.md.
[INDIRECT_PROMPT_INJECTION]: The skill defines an attack surface where instructions could be embedded in the data processed via the Model Context Protocol (MCP).
Ingestion points: Discovery and invocation of story catalogs and live tool schemas (tools/list, tools/call) as detailed in references/automation-and-capture.md.
Boundary markers: None specified for separating control instructions from story metadata or automation parameters.
Capability inventory: Includes application execution (cargo run), UI interaction automation (storybook_run_steps), and file system writes for visual captures (storybook_capture_current_story).
Sanitization: No explicit mention of escaping or validation logic for data received from story-root metadata or external tool outputs.

use-gpui-storybook