The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local shell commands including cargo run, cargo test, git, and gh to perform its bug-fixing workflow. These capabilities are consistent with the skill's primary purpose.
[EXTERNAL_DOWNLOADS]: The installation command in metadata.json retrieves the skill's definition from GitHub (raw.githubusercontent.com), which is a well-known and trusted source for code distribution.
[PROMPT_INJECTION]: The skill processes untrusted data from GitHub issues (gh issue view) and compiler error messages without explicit sanitization or boundary markers, making it vulnerable to indirect prompt injection.
Ingestion points: External GitHub issue descriptions and local Rust error messages (SKILL.md).
Boundary markers: Absent; there are no delimiters separating external data from instructional context.
Capability inventory: Shell execution (cargo run/test), file modification, and repository operations (git push, PR creation) (SKILL.md).
Sanitization: Absent; external data is incorporated directly into the agent's reasoning process.

fix-bug