The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and obey instructions from a specific file path: /tmp/$(basename "$PWD")/code-review.md.
Ingestion points: Reading content from /tmp/<dirname>/code-review.md in Step 2.
Boundary markers: None. The skill is instructed to "fix CRITICAL findings first" from this file without validation or delimiters.
Capability inventory: The skill can execute shell commands (git, perl, rg, go), modify source code, and run build/test suites (go build, go test).
Sanitization: None. The agent treats the content of the external file as a prioritized instruction set for refactoring.
[COMMAND_EXECUTION]: The skill uses perl with the -e flag to perform global text replacements and executes Go toolchain commands (go build, go test, gofmt) to verify changes. While intended for its purpose, this allows for execution of code within the repository's context.

go-refactor