The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references standard package management tools like cabal and stack to download dependencies from the Haskell package repository Hackage.
[REMOTE_CODE_EXECUTION]: The documentation provides examples for installing the GHCup and Stack toolchains using curl piped to sh from get-ghcup.haskell.org and get.haskellstack.org. These are the official and standard installation vectors for the Haskell ecosystem.
[COMMAND_EXECUTION]: Includes instructions for executing shell commands for project initialization, compilation, and dependency management using ghcup, cabal, and stack within a Bash environment.
[INDIRECT_PROMPT_INJECTION]: The skill defines a surface where untrusted user-provided code could potentially influence the agent's actions due to its broad capabilities.
Ingestion points: Haskell code snippets and programming queries processed by the agent (identified in SKILL.md).
Boundary markers: The instructions do not define strict delimiters or warnings to ignore instructions embedded in the code being analyzed.
Capability inventory: The skill environment allows for Bash execution and file modification (Write, Edit) alongside Read operations.
Sanitization: There are no documented mechanisms for sanitizing or isolating untrusted external code before processing.

haskell-expert