The Agent Skills Directory

[SAFE]: The skill documents a legitimate open-source development tool (tsdown) created by the author (oakoss). No evidence of prompt injection, data exfiltration, or persistence mechanisms was found.
[INDIRECT_PROMPT_INJECTION]: The skill operates on local project files, which constitutes a potential ingestion point for instructions if those files are maliciously crafted.
Ingestion points: Metadata and configuration data ingested from package.json, tsconfig.json, and tsdown.config.ts (SKILL.md, configuration.md).
Boundary markers: Standard structured data formats (JSON/TS) serve as delimiters.
Capability inventory: File system writing to dist, execution of build hooks, and potential network access via the plugin system.
Sanitization: Data is validated against internal schemas during the build process.
[DYNAMIC_EXECUTION]: The tool supports programmatic features that execute logic during the build lifecycle.
Logic execution: The onSuccess hook (advanced-features.md) allows for custom asynchronous functions to run after builds.
Configuration loading: The skill uses TypeScript-based configuration files which are dynamically loaded and executed to resolve build settings.

tsdown