The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the host shell and a bundled Python script to perform environment setup and repository analysis.
The instructions direct the agent to create directories using mkdir or New-Item depending on the operating system.
The skill executes a local script scripts/scan.py to aggregate project metadata, identify languages, and find documentation markers.
scripts/scan.py utilizes subprocess.check_output to execute git log and identify high-churn files. The implementation uses a list of arguments and avoids shell=True, following secure coding practices.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from the repository being documented.
Ingestion points: The agent reads the output of the scan.py script (.lcs/codebase/.codebase-scan.txt) and performs direct inspection of source files and manifests during Phase 2.
Boundary markers: No specific delimiters or "ignore instructions" warnings are utilized when the agent interpolates file content into its context.
Capability inventory: The agent has the capability to write files, create directories, and execute the bundled Python analysis script.
Sanitization: No sanitization or filtering of the repository's content is performed before processing. However, the skill explicitly instructs the agent to document only variable names and never expose actual secrets from environment files.

lcs-codebase-doc