The Agent Skills Directory

[COMMAND_EXECUTION]: The orchestration script scripts/analyze_compile.sh uses eval to execute command strings retrieved from the output of scripts/get_compile_command.py. These commands are derived from parsing local Ninja build files.
[DYNAMIC_EXECUTION]: The skill programmatically generates new shell scripts (e.g., compile_single_file_{name}.sh) by interpolating data extracted from the Ninja build system into pre-defined templates. These scripts are subsequently written to the project's output directory.
[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from build artifacts which can influence agent execution.
Ingestion points: The skill ingests data from the build environment via scripts/get_compile_command.py (parsing .ninja files) and scripts/parse_ii.py (parsing .ii preprocessed files).
Boundary markers: The ingestion process lacks explicit boundary markers or security delimiters when processing these external files.
Capability inventory: The skill possesses the capability to execute shell commands (via eval in analyze_compile.sh) and create executable script files on the local file system.
Sanitization: There is no significant sanitization of the content extracted from build files before it is used to construct shell commands, beyond basic handling of Ninja variable escaping.

compile-analysis