The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill presents a vulnerability surface for indirect prompt injection as it is designed to interact with potentially untrusted external data (e.g., package.json files) while having access to powerful system tools.
Ingestion points: The skill ingests data from user requests and file contents accessed through the Read and Grep tools as described in the documentation and examples.
Boundary markers: There are no explicit instructions, delimiters, or safety warnings provided to ensure the agent ignores instructions that might be embedded within the data it is managing.
Capability inventory: The skill is granted access to the Bash(cmd:*), Write, and Edit tools in the frontmatter, allowing for arbitrary shell command execution and file system modifications.
Sanitization: No sanitization, schema validation, or escaping logic is defined in the instructions to protect against malicious payloads in the processed files.

package-json-manager