The Agent Skills Directory

PROMPT_INJECTION (HIGH): The skill establishes a high-risk attack surface for Indirect Prompt Injection by granting the agent write-access tools to process untrusted external content.
Ingestion points: Ingests external source code using Read, Grep, and Glob tools.
Boundary markers: Absent; there are no instructions or delimiters provided to help the agent distinguish between its own instructions and untrusted data from the filesystem.
Capability inventory: The agent is granted Write and Edit permissions, enabling it to perform impactful modifications to the codebase based on ingested data.
Sanitization: Absent; the skill lacks any mechanisms to validate or filter data retrieved from external files.
NO_CODE (INFO): No executable scripts or active code logic were detected within the skill files; the vulnerability arises from the tool permissions and instructional workflow defined in the metadata.

boundary-generator