The Agent Skills Directory

[COMMAND_EXECUTION]: The script build-map.py invokes system utilities git and ctags to list repository files and extract symbols. These calls are implemented using subprocess.run with argument lists, which prevents shell injection vulnerabilities.
[PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted repository data, creating a surface for indirect prompt injection.
Ingestion points: Repository source files are read by the build-map.py script (during dependency and entry point detection) and by the agent itself during the module summarization phase.
Boundary markers: The script uses  and  to separate auto-generated blocks from user-edited summaries in Markdown files, but there are no specific boundary markers or 'ignore' instructions applied when the agent reads raw source code.
Capability inventory: The skill has the capability to execute git and ctags via subprocesses and possesses file system write permissions to the .projectmap/ directory.
Sanitization: No content sanitization or instruction-filtering is performed on the source code before it is presented to or processed by the agent.

project-map