The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill reads the contents of instruction rule files to provide analysis, which creates a surface where malicious instructions in those files could influence the agent's behavior.
Ingestion points: The skill uses the file search and read capabilities to ingest all .md files found in .claude/rules/ and ~/.claude/rules/ (Step 1 and Step 2).
Boundary markers: The instructions do not specify the use of delimiters or 'ignore embedded instructions' prompts when the agent processes the rule content.
Capability inventory: The agent has access to Read, Glob, Bash, and Edit tools, allowing it to modify files based on its evaluation.
Sanitization: There is no instructed validation or sanitization of the markdown content before it is processed by the agent's internal logic.
[COMMAND_EXECUTION]: The skill uses shell commands to gather metadata about the repository environment.
Evidence: Executes git ls-files | wc -l via bash to determine the total number of files in the project, which is used to calculate whether a rule's scope is too broad (Step 1).

eval-rules