The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from multiple untrusted sources, including user-provided arguments, project briefs, and output from other discovery tools. This data is used to generate filenames, fill frontmatter fields, and populate file content, which could be exploited to influence agent behavior if the source data contains malicious instructions.
Ingestion points: User arguments (idea/slice), markdown files under context/foundation/briefs/, and roadmap files under context/efforts/.
Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded commands when processing these external contents.
Capability inventory: The skill has permissions to read and write files within the project directory structure and execute shell commands via git.
Sanitization: There is no evidence of sanitization or escaping of ingested content before it is used in file creation or prompt interpolation.
[COMMAND_EXECUTION]: The skill executes git rev-parse HEAD to retrieve the current commit hash for inclusion in research metadata. This is a restricted and benign use of command execution intended for provenance tracking.

dx-new