The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from local project files which could be manipulated to influence agent behavior.
Ingestion points: SKILL.md instructions specify reading Makefile, package.json, pyproject.toml, Cargo.toml, and go.mod via shell commands and agent tools.
Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded malicious content within these files.
Capability inventory: The skill uses Write, Edit, and TodoWrite tools to modify project files based on the analyzed content.
Sanitization: No explicit sanitization or validation of the content read from external project files is performed before processing.
[DYNAMIC_CONTEXT_INJECTION]: The skill utilizes load-time shell execution to populate its context.
Evidence: Multiple !command instances are used in SKILL.md to run pwd, find, and grep to determine project structure and Makefile contents.
Context: These commands are used for legitimate project discovery and do not involve sensitive path access or user-controlled argument injection at load time.

configure-makefile