The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute various shell commands to manage its environment, including using find and rg (ripgrep) to explore the repository, and running local scripts such as update-repos.sh and install.sh for repository maintenance and installation tasks.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it facilitates the ingestion of external data from the TileLang repository into the agent's context.
Ingestion points: The agent is directed to read documentation (docs/), examples (examples/), and core source files (tilelang/) to derive implementation patterns and API usage, as specified in SKILL.md and quick-reference.md.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to help it distinguish between trusted instructions and untrusted content retrieved from the repository files.
Capability inventory: The skill provides the agent with the ability to execute shell commands and local scripts based on its interpretation of the repository content.
Sanitization: The skill does not implement or describe any sanitization, validation, or filtering mechanisms for the data read from the external source.

tilelang-skill