The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions include various shell commands for repository management (git fetch, git branch, git push), build processes (cargo generate-lockfile, cargo build), and file searching (rg).
[DYNAMIC_EXECUTION]: The skill utilizes a Ruby one-liner (ruby -e) to parse and validate YAML configuration files. The code executed is a static string defined within the skill's workflow instructions.
[INDIRECT_PROMPT_INJECTION]: The skill processes content from the project's codebase to infer versions and update documentation, which creates a potential surface for indirect injection if the repository content is compromised.
Ingestion points: Content is read from Cargo.toml, .github/nightly-alpha-branches.yaml, README.md, and files within the docs/ and fern/ directories.
Boundary markers: No specific delimiters are employed to isolate ingested file content from the agent's primary instructions.
Capability inventory: The skill has the capability to write to remote repositories via git push and execute local task runners like just and cargo.
Sanitization: The skill relies on manual user review of documentation matches before application rather than automated content sanitization.

prepare-code-freeze