The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute git commands (rev-parse, symbolic-ref, diff) for repository discovery and find/ls commands to locate its internal components on the local filesystem.
[DYNAMIC_EXECUTION]: The skill dynamically resolves the path to its core logic script (workflows/improve.js) by searching through plugin root environment variables and local home directories, then executes the resulting path via the Workflow tool.
[INDIRECT_PROMPT_INJECTION]: The skill processes data that may contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Processes branch diffs (code and comments) via git diff and accepts a decision string from the user that is relayed verbatim to the workflow.
Boundary markers: There are no explicit instructions to use delimiters or ignore embedded instructions when interpolating these data sources into the workflow arguments.
Capability inventory: The skill possesses the ability to execute shell commands, read files, and trigger secondary workflows or agents (pr-review-toolkit:review-pr).
Sanitization: No sanitization or validation logic is defined for the external code changes or user-provided decision strings before they are processed by the agent.

pr-improver