The Agent Skills Directory

[PROMPT_INJECTION] (HIGH): Indirect Prompt Injection vulnerability surface. The skill ingests untrusted data from the user's filesystem and possesses high-privilege capabilities.
Ingestion points: The skill reads external project files via the <project-path> and <target-path> arguments in SKILL.md.
Boundary markers: Absent. There are no instructions to the agent to ignore embedded commands or markers to delimit code from instructions.
Capability inventory: The skill executes local Python scripts (project_architect.py) which are described as providing "Automated fixes" (file-write capability) and performs analysis via subprocess calls (python3).
Sanitization: Absent. There is no evidence of validation or filtering of the external content before it influences agent decision-making or file modifications.
[COMMAND_EXECUTION] (MEDIUM): The skill relies on executing multiple local Python scripts (architecture_diagram_generator.py, project_architect.py, dependency_analyzer.py) via the command line. While these are part of the skill, their execution on arbitrary user paths poses a risk if the scripts themselves are compromised or behave unexpectedly with malicious input.
[EXTERNAL_DOWNLOADS] (MEDIUM): The development workflow includes npm install and pip install -r requirements.txt. These represent unverifiable dependencies that could lead to supply chain attacks, as specific versions or trusted registries are not enforced in the provided instructions.

senior-architect