The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill establishes a workflow where the agent ingests data from external sources, creating a potential attack surface.
Ingestion points: The reference.md file directs the agent to query the context7 docs tool and the GitHub CLI (gh) for information from various repositories, including gleam-lang/stdlib, rawhat/mist, and schurhammer/gleamy_structures.
Boundary markers: There are no explicit instructions or delimiters defined to help the agent distinguish between its own system instructions and instructions that might be embedded within the external documentation or code it retrieves.
Capability inventory: The skill provides the agent with the ability to execute code via gleam run and gleam test, and to interact with the GitHub API via the gh CLI. This creates a path where instructions found in external data could influence local code execution.
Sanitization: The skill does not include instructions for sanitizing or validating the content retrieved from these external libraries before processing it.

gleam