The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by consuming external data from specification files (specs/*.md) and project configuration (CLAUDE.md) as authoritative instructions for its implementation logic.
Ingestion points: Files located at specs/*.md and CLAUDE.md within the repository.
Boundary markers: Absent; the skill is explicitly instructed that the 'spec is the source of truth'.
Capability inventory: The skill has the ability to write files, create Git commits, push to remote branches, and execute multiple shell commands via the Rust toolchain (cargo).
Sanitization: No validation or sanitization of the specification content is performed before the agent acts on the instructions.
[COMMAND_EXECUTION]: The skill executes local shell commands including cargo fmt, cargo clippy, cargo nextest, and cargo run. While these are standard for Rust development, they are used to validate and execute code that the agent generates dynamically from potentially untrusted specifications.

implement-spec