The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to read and analyze source code files, which serves as a primary ingestion point for untrusted data.
Ingestion points: The skill accesses project files in directories such as src/**, lib/**, and app/**, specifically targeting .js, .ts, .py, .java, and .go extensions.
Boundary markers: There are no explicit instructions or delimiters defined to help the agent distinguish between its own instructions and natural language instructions that might be embedded within the source code comments or strings of the files being analyzed.
Capability inventory: The agent has access to Read, Grep, Glob, and WebSearch tools, allowing it to process and potentially act on content found within the analyzed files.
Sanitization: No sanitization or validation mechanisms are implemented for the external content before it is interpolated into the agent's context for analysis.
[COMMAND_EXECUTION]: The skill utilizes shell hooks to automate environment discovery during the initialization phase.
Execution method: The pre_execution hook executes a series of shell commands including find, grep, wc, xargs, and ls to identify files and locate configuration files.
Evidence: The command ls -la .eslintrc* .prettierrc* .pylintrc tslint.json 2>$dev$null contains an unconventional redirection (2>$dev$null) which may be a typographical error for /dev/null or an attempt to redirect error output to a variable-controlled path.
[METADATA_POISONING]: The skill's internal metadata contains deceptive information regarding its origin.
Evidence: The metadata field author is set to "Claude Code", which conflicts with the actual author attribution. This represents a deceptive metadata pattern that could lead a user to misjudge the skill's source and safety profile.

agent-analyze-code-quality